teamchong/pxpipePublic

cut Claude Code token usage by rendering text context as images

AI summary: A local proxy pipeline that cuts Claude Code token costs by rendering bulky text context into compact images.

Stars
7.9K
+23 today
Forks
678
Watchers
30
Open issues
31
Open PRs
27
Contributors
~33
Commits
543
Branches
14

TypeScriptMITCreated May 20, 2026Last push 1d agoLatest release v0.16.0+159 stars this week+335 this month

Quick answers

What is pxpipe?
A local proxy pipeline that cuts Claude Code token costs by rendering bulky text context into compact images.
What does pxpipe do?
pxpipe is a local proxy server designed to significantly reduce the API token usage of Claude Code by exploiting pricing differences between text and image inputs. Instead of sending large text blocks like system prompts, tool documentation, and history directly as text, pxpipe intercepts the request and renders these bulky sections into compact PNG images before forwarding them. Since the cost of an image token is determined by its pixel dimensions rather than character count, this dense visual encoding yields significantly lower costs per request. The proxy operates transparently, leveraging the same vision channel that Anthropic uses for computer use capabilities.
Who is pxpipe for?
Heavy users of Claude Code and Anthropic APIs who are looking to optimize their token expenditures. It is built for developers comfortable running local proxy servers to manage their AI interaction costs.
How do I get started with pxpipe?
npm install pxpipe
How popular is pxpipe on GitHub?
teamchong/pxpipe has 7,876 stars and 678 forks on GitHub, and gained 159 stars in the last 7 days.
What license does pxpipe use?
teamchong/pxpipe is released under the MIT license.

Star history

since Jul 29, 2026
02K4K6KJul 2026Sep 2026Oct 2026Nov 2026
7.9K stars as of Nov 20, 2026. Measured daily since Jul 29, 2026; GitHub no longer exposes earlier star timestamps.

Update history

4 recorded
  • Nov 20, 2026New release v0.16.0 (previously v0.15.1).
  • Oct 29, 2026New release v0.15.1 (previously v0.15.0).
  • Oct 18, 2026New release v0.15.0 (previously v0.14.0).
  • Oct 6, 2026New release v0.14.0 (previously v0.13.2).

Contribution activity

commits per day, last 52 weeks
NovDecJanFebMarAprMayJunJulAugSepOctNovMonWedFri2025-11-23: 0 commits2025-11-24: 0 commits2025-11-25: 0 commits2025-11-26: 0 commits2025-11-27: 0 commits2025-11-28: 0 commits2025-11-29: 0 commits2025-11-30: 0 commits2025-12-01: 0 commits2025-12-02: 0 commits2025-12-03: 0 commits2025-12-04: 0 commits2025-12-05: 0 commits2025-12-06: 0 commits2025-12-07: 0 commits2025-12-08: 0 commits2025-12-09: 0 commits2025-12-10: 0 commits2025-12-11: 0 commits2025-12-12: 0 commits2025-12-13: 0 commits2025-12-14: 0 commits2025-12-15: 0 commits2025-12-16: 0 commits2025-12-17: 0 commits2025-12-18: 0 commits2025-12-19: 0 commits2025-12-20: 0 commits2025-12-21: 0 commits2025-12-22: 0 commits2025-12-23: 0 commits2025-12-24: 0 commits2025-12-25: 0 commits2025-12-26: 0 commits2025-12-27: 0 commits2025-12-28: 0 commits2025-12-29: 0 commits2025-12-30: 0 commits2025-12-31: 0 commits2026-01-01: 0 commits2026-01-02: 0 commits2026-01-03: 0 commits2026-01-04: 0 commits2026-01-05: 0 commits2026-01-06: 0 commits2026-01-07: 0 commits2026-01-08: 0 commits2026-01-09: 0 commits2026-01-10: 0 commits2026-01-11: 0 commits2026-01-12: 0 commits2026-01-13: 0 commits2026-01-14: 0 commits2026-01-15: 0 commits2026-01-16: 0 commits2026-01-17: 0 commits2026-01-18: 0 commits2026-01-19: 0 commits2026-01-20: 0 commits2026-01-21: 0 commits2026-01-22: 0 commits2026-01-23: 0 commits2026-01-24: 0 commits2026-01-25: 0 commits2026-01-26: 0 commits2026-01-27: 0 commits2026-01-28: 0 commits2026-01-29: 0 commits2026-01-30: 0 commits2026-01-31: 0 commits2026-02-01: 0 commits2026-02-02: 0 commits2026-02-03: 0 commits2026-02-04: 0 commits2026-02-05: 0 commits2026-02-06: 0 commits2026-02-07: 0 commits2026-02-08: 0 commits2026-02-09: 0 commits2026-02-10: 0 commits2026-02-11: 0 commits2026-02-12: 0 commits2026-02-13: 0 commits2026-02-14: 0 commits2026-02-15: 0 commits2026-02-16: 0 commits2026-02-17: 0 commits2026-02-18: 0 commits2026-02-19: 0 commits2026-02-20: 0 commits2026-02-21: 0 commits2026-02-22: 0 commits2026-02-23: 0 commits2026-02-24: 0 commits2026-02-25: 0 commits2026-02-26: 0 commits2026-02-27: 0 commits2026-02-28: 0 commits2026-03-01: 0 commits2026-03-02: 0 commits2026-03-03: 0 commits2026-03-04: 0 commits2026-03-05: 0 commits2026-03-06: 0 commits2026-03-07: 0 commits2026-03-08: 0 commits2026-03-09: 0 commits2026-03-10: 0 commits2026-03-11: 0 commits2026-03-12: 0 commits2026-03-13: 0 commits2026-03-14: 0 commits2026-03-15: 0 commits2026-03-16: 0 commits2026-03-17: 0 commits2026-03-18: 0 commits2026-03-19: 0 commits2026-03-20: 0 commits2026-03-21: 0 commits2026-03-22: 0 commits2026-03-23: 0 commits2026-03-24: 0 commits2026-03-25: 0 commits2026-03-26: 0 commits2026-03-27: 0 commits2026-03-28: 0 commits2026-03-29: 0 commits2026-03-30: 0 commits2026-03-31: 0 commits2026-04-01: 0 commits2026-04-02: 0 commits2026-04-03: 0 commits2026-04-04: 0 commits2026-04-05: 0 commits2026-04-06: 0 commits2026-04-07: 0 commits2026-04-08: 0 commits2026-04-09: 0 commits2026-04-10: 0 commits2026-04-11: 0 commits2026-04-12: 0 commits2026-04-13: 0 commits2026-04-14: 0 commits2026-04-15: 0 commits2026-04-16: 0 commits2026-04-17: 0 commits2026-04-18: 0 commits2026-04-19: 0 commits2026-04-20: 0 commits2026-04-21: 0 commits2026-04-22: 0 commits2026-04-23: 0 commits2026-04-24: 0 commits2026-04-25: 0 commits2026-04-26: 0 commits2026-04-27: 0 commits2026-04-28: 0 commits2026-04-29: 0 commits2026-04-30: 0 commits2026-05-01: 0 commits2026-05-02: 0 commits2026-05-03: 0 commits2026-05-04: 0 commits2026-05-05: 0 commits2026-05-06: 0 commits2026-05-07: 0 commits2026-05-08: 0 commits2026-05-09: 0 commits2026-05-10: 0 commits2026-05-11: 0 commits2026-05-12: 0 commits2026-05-13: 0 commits2026-05-14: 0 commits2026-05-15: 0 commits2026-05-16: 0 commits2026-05-17: 0 commits2026-05-18: 35 commits2026-05-19: 30 commits2026-05-20: 25 commits2026-05-21: 18 commits2026-05-22: 9 commits2026-05-23: 7 commits2026-05-24: 8 commits2026-05-25: 4 commits2026-05-26: 4 commits2026-05-27: 1 commit2026-05-28: 0 commits2026-05-29: 3 commits2026-05-30: 0 commits2026-05-31: 3 commits2026-06-01: 0 commits2026-06-02: 0 commits2026-06-03: 0 commits2026-06-04: 0 commits2026-06-05: 1 commit2026-06-06: 0 commits2026-06-07: 0 commits2026-06-08: 0 commits2026-06-09: 20 commits2026-06-10: 12 commits2026-06-11: 9 commits2026-06-12: 5 commits2026-06-13: 0 commits2026-06-14: 2 commits2026-06-15: 1 commit2026-06-16: 3 commits2026-06-17: 12 commits2026-06-18: 3 commits2026-06-19: 13 commits2026-06-20: 1 commit2026-06-21: 4 commits2026-06-22: 5 commits2026-06-23: 4 commits2026-06-24: 0 commits2026-06-25: 9 commits2026-06-26: 4 commits2026-06-27: 4 commits2026-06-28: 0 commits2026-06-29: 1 commit2026-06-30: 3 commits2026-07-01: 12 commits2026-07-02: 7 commits2026-07-03: 16 commits2026-07-04: 13 commits2026-07-05: 2 commits2026-07-06: 0 commits2026-07-07: 0 commits2026-07-08: 1 commit2026-07-09: 6 commits2026-07-10: 3 commits2026-07-11: 7 commits2026-07-12: 4 commits2026-07-13: 2 commits2026-07-14: 2 commits2026-07-15: 1 commit2026-07-16: 11 commits2026-07-17: 7 commits2026-07-18: 8 commits2026-07-19: 4 commits2026-07-20: 2 commits2026-07-21: 4 commits2026-07-22: 8 commits2026-07-23: 0 commits2026-07-24: 0 commits2026-07-25: 3 commits2026-07-26: 4 commits2026-07-27: 4 commits2026-07-28: 6 commits2026-07-29: 0 commits2026-07-30: 0 commits2026-07-31: 1 commit2026-08-01: 2 commits2026-08-02: 1 commit2026-08-03: 4 commits2026-08-04: 1 commit2026-08-05: 1 commit2026-08-06: 4 commits2026-08-07: 0 commits2026-08-08: 11 commits2026-08-09: 14 commits2026-08-10: 0 commits2026-08-11: 2 commits2026-08-12: 1 commit2026-08-13: 0 commits2026-08-14: 0 commits2026-08-15: 0 commits2026-08-16: 0 commits2026-08-17: 1 commit2026-08-18: 7 commits2026-08-19: 1 commit2026-08-20: 3 commits2026-08-21: 0 commits2026-08-22: 0 commits2026-08-23: 0 commits2026-08-24: 1 commit2026-08-25: 0 commits2026-08-26: 0 commits2026-08-27: 0 commits2026-08-28: 0 commits2026-08-29: 0 commits2026-08-30: 2 commits2026-08-31: 0 commits2026-09-01: 0 commits2026-09-02: 0 commits2026-09-03: 1 commit2026-09-04: 1 commit2026-09-05: 7 commits2026-09-06: 1 commit2026-09-07: 0 commits2026-09-08: 0 commits2026-09-09: 0 commits2026-09-10: 0 commits2026-09-11: 0 commits2026-09-12: 0 commits2026-09-13: 0 commits2026-09-14: 0 commits2026-09-15: 0 commits2026-09-16: 0 commits2026-09-17: 0 commits2026-09-18: 0 commits2026-09-19: 0 commits2026-09-20: 0 commits2026-09-21: 0 commits2026-09-22: 0 commits2026-09-23: 0 commits2026-09-24: 0 commits2026-09-25: 6 commits2026-09-26: 0 commits2026-09-27: 1 commit2026-09-28: 4 commits2026-09-29: 0 commits2026-09-30: 2 commits2026-10-01: 0 commits2026-10-02: 0 commits2026-10-03: 0 commits2026-10-04: 0 commits2026-10-05: 0 commits2026-10-06: 0 commits2026-10-07: 0 commits2026-10-08: 0 commits2026-10-09: 0 commits2026-10-10: 0 commits2026-10-11: 0 commits2026-10-12: 0 commits2026-10-13: 0 commits2026-10-14: 0 commits2026-10-15: 0 commits2026-10-16: 0 commits2026-10-17: 11 commits2026-10-18: 13 commits2026-10-19: 1 commit2026-10-20: 3 commits2026-10-21: 0 commits2026-10-22: 1 commit2026-10-23: 0 commits2026-10-24: 0 commits2026-10-25: 2 commits2026-10-26: 0 commits2026-10-27: 0 commits2026-10-28: 0 commits2026-10-29: 0 commits2026-10-30: 0 commits2026-10-31: 3 commits2026-11-01: 0 commits2026-11-02: 0 commits2026-11-03: 0 commits2026-11-04: 0 commits2026-11-05: 0 commits2026-11-06: 3 commits2026-11-07: 0 commits2026-11-08: 0 commits2026-11-09: 0 commits2026-11-10: 0 commits2026-11-11: 2 commits2026-11-12: 0 commits2026-11-13: 1 commit2026-11-14: 0 commits2026-11-15: 0 commits2026-11-16: 2 commits2026-11-17: 1 commit2026-11-18: 1 commit2026-11-19: 1 commit2026-11-20: 0 commits2026-11-21: 0 commits
525 commits in the last yearLessMore

Signals and awards

derived from tracked data
  • Very active

    525 commits in 52 weeks

  • Well documented

    High community health score

  • Permissive license

    MIT

  • Continuous integration

    Automated checks passing

  • Repeat trending

    3 trending appearances

What pxpipe does

pxpipe is a local proxy server designed to significantly reduce the API token usage of Claude Code by exploiting pricing differences between text and image inputs. Instead of sending large text blocks like system prompts, tool documentation, and history directly as text, pxpipe intercepts the request and renders these bulky sections into compact PNG images before forwarding them. Since the cost of an image token is determined by its pixel dimensions rather than character count, this dense visual encoding yields significantly lower costs per request. The proxy operates transparently, leveraging the same vision channel that Anthropic uses for computer use capabilities.

Heavy users of Claude Code and Anthropic APIs who are looking to optimize their token expenditures. It is built for developers comfortable running local proxy servers to manage their AI interaction costs.

  • Context image rendering: Dynamically converts bulky system prompts, tool docs, and request history into highly compact PNG images.
  • Transparent local proxying: Operates seamlessly between the local Claude Code client and the upstream Anthropic API.
  • Token usage reduction: Lowers end-to-end API billing by exploiting the fixed pixel-based pricing of vision models over dense text.
  • Event logging: Provides a detailed local JSONL event log for auditing token usage and measuring counterfactual cost savings.
  • Vision channel exploitation: Reliably utilizes Anthropic's existing vision capabilities intended for screenshots to read dense text data.

Where teams use it

Cost reduction for heavy usage

Frequent Claude Code users can drastically lower their Anthropic API bills when processing large codebases or long histories.

Maximizing context windows

Developers can effectively pack more information into the model's context window by bypassing strict text token limits.

Local pipeline observability

Engineers can monitor exact token expenditures and savings per request using the detailed local event logging.

Transparent API intercepting

Users can run the proxy seamlessly in the background without needing to modify the core functionality of Claude Code.

Getting started: npm install pxpipe

README

main branch

pxpipe

Cut coding-agent input tokens by rendering bulky context as images — the same system prompt, tool docs, and history, in a fraction of the tokens.

An image's token cost is fixed by its pixel dimensions, not by how much text is inside it. Dense content (code, JSON, tool output) packs ~3.1 chars per image-token vs ~1 char per text-token on real Claude Code traffic. The reader is the same vision channel that Anthropic's computer use already relies on for screenshots. pxpipe is a local proxy that uses that channel for context: it rewrites the bulky parts of each request into compact PNGs before it leaves your machine. On the original June 2026 claude-fable-5 runs, that landed as a ~59–70% lower end-to-end bill (10/10 SWE-bench Lite, 14/19 SWE-bench Pro, 13/15 dense hex). Since then, Gemini 4 (gemini-4-argon) reached 15/15 on pure-image dense hex (with no factsheet in the benchmark), while later Claude releases (claude-opus-5-5, claude-sonnet-5-5, claude-haiku-5-5) regressed sharply on dense pixel reading (0/15–3/15 hex; claude-fable-5-1 and claude-fable-5-2 recover to 11/15–14/15 with +2px row spacing; see The honest part). Durable token savings are measured per-request against a free count_tokens counterfactual in ~/.pxpipe/events.jsonl.

This is what the model sees instead of text:

example: a real transformRequest output: system prompt + tool docs reflowed into one dense page, instruction banner on top, ↵ marking original newlines

~48k chars of system prompt + tool docs: ≈25k tokens as text, ≈2.7k image tokens as this page. Real pipeline output; Gemini 4 Argon (15/15 pure-image hex, 100/100 pure-image arithmetic), Gemini 3.6–3.8 Flash (14/15 hex), and original Fable 5 (13/15 hex) read renders like this at full fidelity, whereas later Claude 5.5 releases do not and Fable 5.1/5.2 need +2px row spacing (see model tiers).

chart: characters a frontier context window holds, 2018–2026 — vendor text series including Grok 4.5; orange measured overlays are Fable 5 [1m] + pxpipe ~19.0M (4.8×) and Gemini 3.6 Flash + pxpipe ~21.3M (5.3×)

Eight years of context growth, in characters. Every text line tops out near ~4M chars (a 1M-token window at ~4 chars/token); Grok 4.5 is shown as a text-window point only (500K). The orange overlays are the same 1M windows read through pxpipe images — ~19.0M chars for Fable 5 (4.8×) and ~21.3M chars for Gemini 3.6 Flash (5.3× text capacity). Density is measured from a live render at generation time, not hand-typed: regenerate with npx tsx scripts/gen-context-chart.ts (source).

Demo

claude-fable-5 (June 2026 release, 13/15 hex / 100/100 arithmetic reader) — plain left, pxpipe right:

Fable-AB-Demo.mp4

On original claude-fable-5, pxpipe counts an exact token 10/10 across 39 imaged filler files (matches grep line-for-line), gets the multi-step ledger arithmetic right, and ends the session at $6.06 with context to spare (73.5k/1M) vs $42.21 at 96% full. Note: these headline Claude numbers are from claude-fable-5, not claude-opus-5-5, claude-sonnet-5-5, or claude-haiku-5-5 (which regressed to 0/15–3/15 on dense pixel reading). One caveat visible in the clip: the pxpipe arm needed a nudge to match the requested one-line output format.

Try it (30 seconds)

npx pxpipe-proxy                                  # proxy on 127.0.0.1:47821
pxpipe warp -- agy                                # Gemini 4 / Antigravity CLI (15/15 pure-image reader)
pxpipe warp -- omp                                # omp / OpenCode / Claude Code (pxpipe warp -- claude)

Dashboard at http://127.0.0.1:47821/: tokens saved, every text→image conversion side by side, kill switch, live model chips. Responses stream normally — pxpipe compresses the request only, never the model's output. Recent turns stay text; the system prompt, tool docs, and older bulk history are imaged.

pxpipe warp

pxpipe warp -- agy             # Antigravity CLI (Gemini 4 / 3.8 Flash via CloudCode PA)
pxpipe warp -- omp             # omp (Anthropic Messages, OpenAI Responses, or Google CloudCode PA)
pxpipe warp -- opencode        # OpenCode (or: claude, codex, cursor-agent, or any shell alias)

Zero base-URL overrides (ANTHROPIC_BASE_URL / GEMINI_API_BASE stay unset), so agy OAuth, Claude Code /remote-control, claude.ai connectors, and first-party gates keep working. Full instructions in the dashboard.

Built-in default routes divert:

  • api.anthropic.com/v1/messages* (Claude Code)
  • daily-cloudcode-pa.googleapis.com/v1internal:streamGenerateContent* and :generateContent* (agy / Antigravity CLI)
  • generativelanguage.googleapis.com/v1beta/models/*:streamGenerateContent* and :generateContent* (Gemini CLI / Google AI Studio)

When agy or a Gemini client runs Gemini 4 (gemini-4-neon, gemini-4-carbon, gemini-4-hydrogen, gemini-4-argon, gemini-4-pro, gemini-4-flash), pxpipe automatically selects GEMINI_PROFILE: 312-col Spleen 5×8, 96-page history cap, and a compact factsheet alongside a vision encoder that reads dense 5×8 pages at 15/15 verbatim hex and 100/100 arithmetic directly from pixels.

Agents that reach their provider over a custom or local gateway base URL (such as omp or opencode using an OpenAI/Cloudflare gateway, or codex on a local port) can either point their provider baseURL at http://127.0.0.1:47821 or pass a --route rule (a rule that names a port matches only that port):

pxpipe warp --route '127.0.0.1:9090/v1/*=http://127.0.0.1:47821' -- codex

Offline export (no proxy)

You can render text, files, or diffs to PNG pages without running the proxy or connecting Claude Code:

npx pxpipe-proxy export src/
cat prompt.txt | npx pxpipe-proxy export --stdin
npx pxpipe-proxy export --git

If the package is installed, use pxpipe export instead of npx pxpipe-proxy export.

Each run writes a fresh pxpipe-export-XXXXXX/ output folder (the exact path is printed when the command finishes) containing page-*.png, factsheet.txt, manifest.json, and prompt.txt. Upload the PNG pages and paste the prompt into image-upload clients such as Cursor when you want dense visual context without running the proxy.

The honest part

Not every vision encoder can read dense 5×8 bitmap glyphs. Gemini 4 (gemini-4-argon) is the first model to score 15/15 on pure-image dense hex alongside 100/100 pure-image arithmetic, 98/98 gist recall, 18/18 state tracking, and 0/16 confabulations. By contrast, every post-Fable-5 Claude release (claude-opus-5, claude-opus-5-5, claude-sonnet-5-5, claude-haiku-5-5) regressed sharply on dense pixel reading (0/15 to 3/15 on exact hex). claude-fable-5-1 and claude-fable-5-2 read 6/15 and 7/15 on the unspaced cell and 14/15 and 13/15 with +2px row spacing, which is now their default.

tier models dense hex (pure image, N=15) gist (N=98) arithmetic (N=100) runtime status & guidance
Tier 1 — Full-fidelity readers gemini-4-neon 14/15 (15/15 w/ factsheet) 95/98 100/100 On by default. 96-page history cap + compact factsheet; 15/15 dense hex with shipped factsheet.
Tier 1 - Full-fidelity readers gemini-4-carbon 14/15 (15/15 w/ factsheet) 96/98 100/100 On by default. 96-page history cap + compact factsheet; 15/15 dense hex with shipped factsheet.
Tier 1 — Full-fidelity readers gemini-4-argon (gemini-4*) 15/15 98/98 100/100 On by default. First model to hit 15/15 hex and 100/100 arithmetic on pure images; 96-page history cap + compact factsheet.
Tier 1 — Full-fidelity readers google/gemini-3.6-flash, 3.7-flash, 3.8-flash 14/15 98/98 100/100 On by default (compact factsheet). ~5.0–5.3× text capacity.
Tier 1 — Full-fidelity readers claude-fable-5 (original June 2026) 13/15 98/98 100/100 On by default. SWE-bench Lite 10/10, SWE-bench Pro 14/19.
Tier 1 — Full-fidelity readers claude-fable-5-1, claude-fable-5-2 (+2px spaced rows) 14/15 / 13/15 98/98 100/100 On by default (spaced profile, ~25% more image tokens than bare 5×8; 5-2 scored 13/15 and 11/15 across two runs).
Tier 2 — Gist & agentic history only (not byte-exact on dense logs/JSON) claude-opus-5-5 (default), claude-opus-5 (opt-in), grok-4.5 / grok-4.6 (opt-in), gpt-5.6-sol (opt-in) 0/15–3/15 83/98–97/98 98/100–100/100 Do not use for tool-less document/log extraction. Safe only in live agent sessions where recent turns stay text and the agent has tools (Read/Grep) to re-fetch exact IDs/numbers.
Unsupported — Do not image claude-haiku-5-5 (claude-haiku-*), claude-sonnet-5-5, claude-opus-4-8, gpt-6.1-sol, grok-4.7, gpt-6-astra, gpt-6.2-astra, gpt-6.2-sol, kimi-k3, glm-5.3-flash, deepseek-v4.1-flash-1017 0/15–7/15 14/98–98/98 36/100–100/100 Not in default scope (or loses money after cache discount). Weak readers guess plausible wrong numbers/IDs or misbind factsheet tokens when tools are disabled.
  • Independent 6-document benchmark (TokenSave / johnisking corpus — English/Korean/Japanese prose, JS code, 60-record JSON, 230-line log): That benchmark tested only claude-haiku-5-5, claude-sonnet-5-5, claude-opus-5-5, and claude-fable-5-1 with --tools "", where images only scored 3/60 to 14/60 on planted hex IDs, 6/60 to 29/60 on planted numbers, 0/80 on unplanted log/JSON IDs, and 3/10 on Korean prose numbers (94392 misread as 9392). Running the exact same 6-document, 80-item benchmark on gemini-4-argon pure images (no factsheet) scores 28/30 (93.3%) on planted hex IDs (10/10 on English + Code, 10/10 on Korean prose_ko including 94392 exact, 5/5 on Japanese hex), 27/30 (90.0%) on planted numbers, and 13/20 on unplanted JSON/log items (including 4/5 on unplanted 230-line log req= IDs where all four Claude models scored 0/20).
  • We tested removing the factsheet on Gemini 4 (68/80 pure-image) — and put it back: Because gemini-4-argon scored 15/15 on our 12-char verbatim-hex benchmark and 100/100 on pure-image arithmetic, we tested removing the text factsheet completely (factSheetFormat: 'none') on the 6-document TokenSave corpus. Even at 85% overall (28/30 planted hex, 27/30 planted numbers), 5 of the 12 pure-image misses were single-pixel 0↔8 / 6↔8 substitutions inside 22-char req_<18-hex> strings (req_709b2b8ff97e92994a → req_709b2b0ff97e92994a) or 30141 → 38141. In Spleen 5×8, 0 and 8 differ by one interior pixel. When a strong reader (gemini-4-argon or claude-fable-5) reads the image, it already sees which ticket/row goes with which value and uses the bounded 96-token factsheet to disambiguate 0 vs 8 (recovering the planted misses to 30/30 hex and 29/30 numbers), so we kept the factsheet in production. On weak readers (opus-5-5, sonnet-5-5, and unspaced fable-5-1), the model cannot read the image rows at all, so it guesses pairings from the factsheet and fails 0/80 on any ID past the 96-token cap (FINDINGS.md, 2026-10-25 entry).
  • Escape hatch: subagents on non-allowlisted models pass through as text — route byte-exact work there (CLAUDE_CODE_SUBAGENT_MODEL=claude-sonnet-4-6, or model: sonnet in agent frontmatter), or restrict PXPIPE_MODELS=gemini,claude-fable-5 if you do not want claude-opus-5-5 imaged.
  • Real work (measured on claude-fable-5): SWE-bench Lite pilot 10/10 both arms at −65% request size; SWE-bench Pro 14/19 ON vs 15/19 OFF at −60%, verdicts agree 18/19, and the single split re-resolved 3/3 on replication — run-to-run variance, not compression. Small n; receipts in eval/. Later Claude releases (opus-5-5, sonnet-5-5, and fable-5-1/5-2 on bare 5×8) have weaker dense-cell pixel recall than claude-fable-5; fable-5-1/5-2 recover with +2px row spacing.
  • Workload-dependent. Wins on token-dense content (~1 char/token), loses money on sparse prose (~3.5 chars/token); a profitability gate (calibrated on N=391 production rows) images only where the math wins.
  • Client-dependent. Savings track uncached bulk the client still re-sends as text. Claude Code re-sends system + tools + history on /anthropic/messages and typically lands ~60–70%. Details and measured splits: docs/CACHING_AND_SAVINGS.md.
Configuration, `PXPIPE_MODELS`, and per-model rendering details
  • Model scope: default PXPIPE_MODELS=claude-fable-5,claude-opus-5-5,gemini. The gemini base covers every Gemini id (3.6/3.7/3.8 Flash, Pro, 4, 5, and future versions); to opt Gemini out, drop gemini from PXPIPE_MODELS or click the chip off. Opus 5, Sonnet 5.5, Sol, GPT 5.5, and Grok are opt-in only (dashboard chips or PXPIPE_MODELS). Haiku is never enabled. The exact Sol id still matters: sibling variants such as gpt-5.6-terra do not inherit Sol's allowlist or render profile. PXPIPE_MODELS=off disables imaging. Everything else passes through byte-identical. On the GPT path, tool definitions stay native JSON and no Anthropic cache_control markers are used. Responses history compression recognizes completed function_call/function_call_output pairs, including OpenCode's parallel calls-then-outputs rounds: only old closed rounds are imaged atomically; every open call and malformed/orphan state remains native. The base profile keeps the newest six completed pairs and allows 32 images; Sol keeps one pair and allows 64 images, while Grok allows 24 images. Opt-in long-session coverage can be changed (defensive cap 100) with PXPIPE_GPT_HISTORY_MAX_IMAGES=48 after validating the provider's request cap.
  • claude-opus-5-5 & claude-opus-5: claude-opus-5-5 is on by default at spaced 5×8 (+2px row height, ~25% more image tokens than Fable's layout, still ~50%+ live savings). Good for gist/reasoning (94/98 gist, 100/100 arithmetic, 0/16 never-stated), weak on dense verbatim hex (3/15 on Opus 5.5, 2/15 on Opus 5); agents must re-fetch exact IDs via tools. Details: FINDINGS.md.
  • Per-model rendering: Gemini 4 (GEMINI_PROFILE) uses 312-column, 1568×728 5×8 Spleen with 96 max history pages, mixed history grouping, and a compact factsheet; Gemini 3.x uses the same 5×8 canvas with a 32-page cap. Claude uses 312-column 5×8 Spleen (dense for original Fable 5, +2px row spacing for Fable 5.1/5.2 and Opus/Sonnet). Opt-in gpt-5.6-sol and Grok use native 14px JetBrains Mono glyphs in a 9×16 cell, 84 columns, and a 764px full-width strip. Sol receipts and profile evidence.
  • Grok 4.5 / 4.6 (opt-in): native 14px / 84 cols / maxH 512 (100/100 arith, 97/98 gist). Off by default (dense hex still 0/15). History uses mixed collapse so Codex assistant messages between tool rounds still image. Enable with PXPIPE_MODELS=claude-fable-5,grok-4.6 or the dashboard chip. eval/grok-density/QUALITY_RESULTS.md.

Benchmark results and receipts

Model quality

This matrix shows coverage as well as scores. — means the model was not run on that test; it does not mean zero. Every dense hex (N=15) score is a pure-image read with NO factsheet (12-char hex IDs looked up by dur_ms across 80-line JSON event-log PNGs). Arithmetic uses novel random-number problems. Gist, state, and never-stated probes share one corpus. Never-stated is confabulations, so lower is better. The numbers at column is the render geometry the row's scores were measured at; a model's shipped profile can differ (Sol and Qwen ship the measured 14px/84 geometry, but their broad-suite numbers predate it).

model numbers at arithmetic (N=100) gist (N=98) state (N=18) never-stated (N=16) dense hex, pure image (N=15) profile provenance and receipts
gemini-4-neon (xhigh via agy) Spleen 5×8, 312 cols (shipped) 100/100 95/98 18/18 0/16 14/15 profile-aligned v2 suite (100/100 pure-image arithmetic; 15/15 hex with compact factsheet, 14/15 pure-image): quality results
gemini-4-carbon (default) Spleen 5×8, 312 cols (shipped) 100/100 96/98 18/18 0/16 14/15 profile-aligned v2 suite (100/100 pure-image arithmetic; 15/15 hex with compact factsheet, 14/15 pure-image): quality results
gemini-4-hydrogen (xhigh via agy) Spleen 5×8, 312 cols (shipped) 100/100 95/98 18/18 0/16 — profile-aligned v2 suite (100/100 pure-image arithmetic; 15/15 hex with compact factsheet): quality results
gemini-4-argon (default) Spleen 5×8, 312 cols (shipped) 100/100 98/98 18/18 0/16 15/15 shipped Gemini 4 profile; 100/100 pure-image arithmetic & 15/15 pure-image hex: quality results
google/gemini-3.6-flash, 3.7-flash, 3.8-flash (default) Spleen 5×8, 312 cols (shipped) 100/100 98/98 18/18 0/16 14/15 current shipped Gemini 3.x profile: quality results
claude-fable-5 (default) Spleen 5×8, 312 cols (shipped) 100/100 98/98 18/18 0/16 13/15 June 2026 production profiles: arithmetic + hex, gist/state/guards
claude-fable-5-1 (default) Spleen 5×8, 312 cols, +2px rows 100/100 98/98 18/18 0/16 14/15 spaced profile; was 6/15 hex and 95/98 gist on the unspaced Fable 5 cell. Arithmetic measured unspaced, not rerun: arithmetic, dense hex, gist/state/guards
claude-fable-5-2 (default) Spleen 5×8, 312 cols, +2px rows 100/100 98/98 18/18 0/16 13/15 spaced profile; 13/15 and 11/15 over two hex runs, was 7/15 and 5/15 on the unspaced Fable 5 cell. Arithmetic measured unspaced, not rerun: arithmetic, dense hex, gist/state/guards
claude-opus-5-5 (default, gist-only) Spleen 5×8, 312 cols, +2px rows 100/100 94/98 17/18 0/16 3/15 spaced profile (Opus 5.5+); hex measured at dense geometry: arithmetic, gist/state/guards, dense hex
claude-opus-5 (opt-in) Spleen 5×8, 312 cols (shipped) 100/100 94/98 17/18 0/16 2/15 current profile: arithmetic, gist/state/guards, dense hex
claude-sonnet-5-5 (opt-in, not recommended) Spleen 5×8, 312 cols, +2px rows 100/100 85/98 12/18 0/16 2/15 spaced profile, opt-in via PXPIPE_MODELS; 6 state misses are wrong final values/counts, 7 gist misses answered UNKNOWN: arithmetic, gist/state/guards, dense hex
grok-4.5 (opt-in) JetBrains Mono 14px, 84 cols (shipped) 100/100 97/98 17/18 0/16 0/15 native 14px/84 quality suite (live profile); quality, native-sweep
grok-4.6 high (opt-in) JetBrains Mono 14px, 84 cols (shipped) 100/100 97/98 17/18 0/16 0/15 native 14px/84, reasoning high; quality
gpt-5.6-sol (opt-in) Spleen 5×8, 152 cols; ships 14px/84 98/100 83/98 17/18 4/16 0/15 broad suite predates the shipped 14px profile; 14px pilot: 7/8 exact, 0 inventions, gist/guard pass: pilot
claude-opus-4-8 (historical) Spleen 5×8, 312 cols (historical) 93/100 77/98 18/18 0/16 0/15 historical profile: arithmetic, gist/state/guards, dense hex
moonshotai/kimi-k3 (unsupported) Spleen 5×8, 152 cols (generic default) 79/100 84/98 15/18 1/16 0/15 generic GPT profile, no measured geometry of its own: quality results
qwen-3.8 (@cf/qwen/qwen3.8-27b, opt-in) Spleen 5×8, 152 cols; ships 14px/84 98/100 72/98 11/18 0/16 0/15 broad suite predates the shipped 14px profile; 14px pilot: 8/8 exact, 0 inventions, 11/15 hex: pilot & quality
glm-5.3-flash (@cf/zai-org/glm-5.3-flash, unsupported) Spleen 5×8, 152 cols (default fallback, nothing shipped) 36/100 57/98 6/18 0/16 0/15 5×8 is illegible to GLM (0/15 hex); 14px pilot: 10/15 hex, all misses single-glyph confabs, guards 0/16: pilot & quality
deepseek-v4.1-flash-1017 (@cf/deepseek-ai/deepseek-v4.1-flash-1017, unsupported) JetBrains Mono 14px, 84 cols (pilot, nothing shipped) 100/100 98/98 18/18 0/16 0/15 5×8 is illegible to DeepSeek V4.1 (0/15 hex); 14px pilot: 100/100 arith, 98/98 gist, 18/18 state, 0/16 guards, 8/15 hex: pilot & quality
gpt-6.1-sol (opt-in) Spleen 5×8, 152 cols; ships 14px/84 100/100 20/98 9/18 0/16 0/15 5×8 collapses on dense history (20/98 gist); 14px/84 suite: 100/100 arith, 96/98 gist, 18/18 state, 15/15 hex with factsheet: 5×8, 14px
grok-4.7 high (opt-in) Spleen 5×8, 152 cols; ships 14px/84 85/100 14/98 5/18 0/16 0/15 5×8 collapses on dense history (14/98 gist); 14px/84 suite: 100/100 arith, 98/98 gist, 18/18 state, 14/15 hex with factsheet: 5×8, 14px
gpt-6-astra low (unsupported) Spleen 5×8, 152 cols (original detail) 97/100 26/98 7/18 0/16 2/15 v2 aligned suite (high detail: 98/100 arith, 25/98 gist, 6/18 state, 5/15 hex): quality results
gpt-6.2-astra low (unsupported) Spleen 5×8, 152 cols (original detail) 98/100 24/98 6/18 3/16 7/15 v2 aligned suite (237 calls); weak gist recall and 3 never-stated inventions: quality results
gpt-6.2-sol low (unsupported) Spleen 5×8, 152 cols (original detail) 100/100 40/98 9/18 0/16 7/15 v2 aligned suite on 5×8 fallback (237 calls); weak dense gist recall (40/98): quality results

Native-profile cost check

Offline export of the same deterministic 454,045-character dense record corpus through each complete profile produced:

model profile pages text estimate image tokens savings
Claude, Spleen 5×8 17 122,715 23,856 80.6%
Sol, JetBrains Mono 14px 45 122,715 65,424 46.7%

The text estimate uses 3.7 characters/token; image tokens use each model's provider formula and actual rendered page dimensions. These figures establish profile cost on this corpus, not a universal workload savings rate. Sol's paid fixtures estimated 42% while reading 7/8 exact with no unsupported inventions.

The runs use different transports and profile generations, not one identical image geometry. Fable and Opus use Claude; Gemini 3.x, Argon, and Carbon use Google AI Studio while Gemini 4 Hydrogen and Neon use agy CLI (AGY_CLI=1); Sol and Grok use Codex Responses; Kimi K3 uses Cloudflare's OpenAI-compatible transport. All dense hex (N=15) scores are pure-image reads with no factsheet. Current production profiles include the adjacent bounded factsheet on runtime requests; historical or pure-image exceptions are identified in the linked evaluation.

Model-specific evaluations

These are not cross-model comparisons. Every unlisted model is not run.

test model result evaluation and receipts
SWE-bench Lite claude-fable-5 pxpipe 10/10; text 10/10; −65% request size paired pilot
SWE-bench Pro claude-fable-5 pxpipe 14/19; text 15/19; −60% request size paired pilot
production-history row localization google/gemini-3.6-flash text 17/30; pxpipe 18/30 positional retrieval
production-history exact row google/gemini-3.6-flash text 3/30; pxpipe 3/30 positional retrieval

The SWE-bench runner is Claude Code/Fable-specific; no other model has an ON/OFF run. Gemini's positional-retrieval sweep is directional evidence, not a general Lost-in-the-Middle result.

Capacity / density (how many chars per vision-token?)

Measured by rendering this repo’s dense fixture through the real pipeline and pricing pixels at each family’s vision rate. Multiplier = measured chars/vision-token ÷ 4 (prose text baseline). Not a model-quality score.

family window as text (@4 c/tok) as pxpipe images density multiplier
claude-fable-5[1m] (default) 1M ~4.0M ~18.9M ~18.9 c/vt (exact 28px patches) ~4.7×
google/gemini-3.6-flash 1M ~4.0M ~20.1M ~20.1 c/vt (1,078 tok/page) ~5.0×
claude-opus-5 1M ~4.0M ~18.9M ~18.9 c/vt (resolves to Fable 5’s geometry) ~4.7×

Regenerate: npx tsx scripts/gen-context-chart.ts · chart PNG docs/assets/context-window-chars.png.

The older GSM8K result is omitted because its training-data contamination can hide image misreads; the linked arithmetic evaluations use novel numbers.

How it works

model id ──► render profile ──► wrap/reflow bulk context ──► PNG[] + bounded factsheet

The proxy handles Anthropic Messages, OpenAI Responses and Chat Completions, and Google generateContent requests. It rewrites eligible bulk into image blocks and forwards the provider-native request, or bridges Anthropic Messages to a configured OpenAI-compatible provider. On Anthropic, the static prefix and prompt-cache boundary are preserved. Model-specific profiles control geometry, factsheets, history retention, and profitability, so sparse prose stays text. Events log to ~/.pxpipe/events.jsonl.

Library use (no proxy)

import { renderTextToImages, transformAnthropicMessages } from "pxpipe-proxy";

const { pages } = await renderTextToImages(toolResultText);     // pages[i].png: Uint8Array
const { body, applied, info } = await transformAnthropicMessages({
  body: requestBytes,
  model: "claude-fable-5",
});

options.keepSharp(block) pins blocks as text; options.emitRecoverable returns the originals of imaged blocks. Pure-JS runtime (Node and edge/Workers); @napi-rs/canvas is build-time only. Full API: src/core/index.ts.

Offline stats (no proxy): pxpipe stats

The live dashboard shows savings while the proxy is running. To read the same event log after the fact — with no server up — summarize it straight from disk:

pxpipe stats                   # human report from ~/.pxpipe/events.jsonl
pxpipe stats --json            # same aggregate as machine-readable JSON
pxpipe stats --file /path/to/events.jsonl

Alongside request counts, compression ratios, latency percentiles, and cache-hit rates, the report prints a measured savings headline — count_tokens of the original body versus real usage, over probe-measured rows only (unmeasured requests are excluded, never counted as zero). This is a raw-token figure (cache reads at face value, not cost-weighted), so it is deliberately a different quantity from the dashboard's cost-weighted saved %. Point it at a non-default log with --file, or set PXPIPE_LOG.

Exit codes: 0 report printed, 1 events file not found, 2 file present but no valid events. pxpipe stats --help prints usage.

Development

pnpm install && pnpm test
pnpm run build                # regenerates dist/

Windows is community-supported: primary development targets macOS/Linux, and Windows-specific fixes rely on contributor PRs (thanks @makoribrian).

FAQ

Is the headline end-to-end, or only on the requests you touched?

End-to-end, the whole bill. Most compression tools report savings only on the input slice they touched, which flatters the number. The end-to-end denominator is every production request: the small ones pxpipe correctly left untouched, all cache writes and reads, and all output tokens (which the proxy never compresses). On a 13,709-request snapshot that was 59% ($100 → ~$41); a later 8,904-compressed-request trace measured ~70%. Compressed-only runs higher (~72–74%) and is quoted separately, never as the headline. The exact figure is workload-dependent — reproduce it on your own log.

How is the math measured?

Both sides of the same request, at the same moment. For every /v1/messages POST the proxy fires a free count_tokens probe on the original uncompressed body (the counterfactual) in parallel with the real forward, and reads Anthropic's actually-billed usage block off the response. Both land in the same row of ~/.pxpipe/events.jsonl, so there is no turn-count or run-to-run confound. Dollar conversion uses Fable 5 list ratios: input ×1.0, cache write ×1.25, cache read ×0.1, output ×5. Cache pricing is applied identically to both sides, so the caching discount cancels and cannot be double-counted as "savings". Re-derive it yourself from the events log: the formula and field names are documented in src/core/baseline.ts.

What does it actually compress?

Three kinds of input blocks, each behind a profitability gate:

  1. large tool_result bodies (file reads, command output, logs) above ~6k chars of token-dense content
  2. older collapsed history: turns behind the live tail get re-rendered as image pages, recent turns always stay text
  3. the static cacheable system prompt + tool docs slab; appended non-cacheable system blocks stay live text so host custom instructions keep system-level salience

Everything else passes through byte-identical: your messages, recent turns, the model's output (it is the response, the proxy never touches it), sparse prose, and anything too small to win. Model defaults and detailed results are listed under model support and benchmarks.

Has it ever failed for real, outside the benchmarks?

Yes, once in weeks of daily use: the model recalled a person's name from imaged chat history and got it confidently wrong. No error, just a plausible wrong name. That is the documented failure mode: exact strings in imaged content are not byte-safe. Coding sessions tolerate this because the agent re-reads files before editing; pure chat recall has no such check. This failure mode is measured, not anecdotal: the legibility audit quantifies exact-string recall off rendered pages (blind reads top out at 63% on dense identifiers, with every miss predicted by a glyph-confusability matrix) and documents the shipped mitigations — page geometry clamped to the API's resample cap so billed pixels actually reach the vision encoder, and selected identifiers (SHAs, numbers) riding alongside as text.

Why are misses silent confabulations instead of read errors?

Because model vision is not OCR: the image becomes patch embeddings, never discrete characters, so there is no per-glyph confidence to fail loudly on. When pixels underdetermine a glyph, the language prior fills the gap with something plausible. Mechanism and receipts: docs/NOT-OCR.md.

Didn't DeepSeek-OCR show this doesn't hold up in practice?

No: it proved the channel works, using an encoder/decoder pair trained for the job. The skepticism dates from October 2025, when no stock production model could read dense renders; that changed with June 2026 claude-fable-5 (13/15 pure-image verbatim hex vs 0/15 on prior Opus) and Gemini (14/15 on Gemini 3.6–3.8 Flash, 15/15 on Gemini 4 Argon pure-image), even as later Claude 5.5 and Fable 5.1 releases regressed to 2/15–6/15. Timeline and per-model numbers: docs/NOT-OCR.md.

Why does the README read like an AI wrote it?

Because one did. Most of this repo's commits — the code and the docs — were authored by Opus/Fable agent sessions running behind pxpipe itself, reading their own collapsed history as image pages while they worked.

Additional limitations

  • PNG encoding adds latency to large requests before they leave.
  • ASCII/Latin-1 well tested; CJK works but conservatively.

Research status

Current as of October 2026. Exact recall at 5×8 density is bottlenecked by the model's vision encoder, not just font tweaks: repainting K reduced Fable 5's H/K confusion from 47.2% to 18.7% (FINDINGS.md, 2026-07-19), while later Claude releases (claude-fable-5-1 at 6/15, claude-fable-5-2 at 7/15 on the same cell, 14/15 and 13/15 with +2px rows; claude-opus-5-5 at 3/15, claude-sonnet-5-5 at 2/15) regressed relative to June 2026 claude-fable-5 (13/15). Conversely, Gemini 4 (gemini-4-argon) cleared the 5×8 resolution tripwire at 15/15 pure-image dense hex, 100/100 pure-image arithmetic, and 93% / 90% pure-image hex/number recall on the 6-document TokenSave corpus; we tested removing the factsheet on Gemini 4 and kept it after observing single-pixel 0↔8 substitutions in 22-char req_ IDs (FINDINGS.md, 2026-10-25 entry).

Effective-context benefits remain unproven. The production-history results above are directional evidence, not a general context-window or long-task accuracy claim.

Community projects

Third-party projects listed here are not maintained or supported by pxpipe.

  • pxpipe-windows — Windows support for pxpipe mitm (node-forge CA in place of openssl, Task Scheduler autostart).
  • OmniGlyph — A community-maintained project derived from pxpipe and used by OmniRoute.
  • pxpipe-go — A Go port of pxpipe's core with a CLI wrapper, standalone proxy, and embeddable library for Anthropic Messages and OpenAI Chat/Responses.

License

MIT.

View on GitHub

Recent activity

commits and pull requests

Recent open issues

view all

Discussions

all 4

Releases and announcements

12 total
  1. v0.16.0v0.16.0Nov 11, 2026

    ## What's Changed * fix(google): pin active user prompt and mid-work interjection across history collapse by @teamchong in https://github.com/teamchong/pxpipe/pull/341 * feat(codex): experimental GPT-6 setup and profile-aligned evals by @teamchong in https://github.com/teamchong/pxpipe/pull/273 * fix(proxy): pass upstream streams through without default proxy timeouts by @teamchong in https://github.com/teamchong/pxpipe/pull/347 * perf(node,warp): flush response headers immediately and enable TCP_NODELAY on proxy sockets by @teamchong in https://github.com/teamchong/pxpipe/pull/348 * fix(warp,proxy): propagate mid-stream upstream aborts and retain stream_error receipts by @teamchong in https://github.com/teamchong/pxpipe/pull/349 * eval(fable): add claude-fable-5-2 results by @teamchong in https://github.com/teamchong/pxpipe/pull/351 * release: 0.16.0 by @teamchong in https://github.com/teamchong/pxpipe/pull/352 **Full Changelog**: https://github.com/teamchong/pxpipe/compare/v0.15.1...v0.16.0

  2. v0.15.1v0.15.1Oct 26, 2026

    ## What's Changed * eval(deepseek): add DeepSeek V4.1 Flash 1017 quality benchmarks and 14px pilot by @teamchong in https://github.com/teamchong/pxpipe/pull/329 * fix(warp): match a default port in absolute form, as the CONNECT path does by @dex0shubham in https://github.com/teamchong/pxpipe/pull/334 * fix(transform,openai): forward malformed messages untouched instead of 502ing by @dex0shubham in https://github.com/teamchong/pxpipe/pull/326 * fix(savings): honest estimates under server-side threads and warp by @LZong-tw in https://github.com/teamchong/pxpipe/pull/328 * docs(readme,gemini): clarify model support tiers, Fable 5 baseline, and Gemini 4 15/15 eval by @teamchong in https://github.com/teamchong/pxpipe/pull/335 * release: 0.15.1 by @teamchong in https://github.com/teamchong/pxpipe/pull/336 **Full Changelog**: https://github.com/teamchong/pxpipe/compare/v0.15.0...v0.15.1

  3. v0.15.0v0.15.0Oct 18, 2026

    ## What's Changed * eval: add Claude Sonnet 5.5 results by @teamchong in https://github.com/teamchong/pxpipe/pull/298 * feat(gemini): add gemini-4-argon support and quality evaluation results by @teamchong in https://github.com/teamchong/pxpipe/pull/322 * feat(warp,proxy): route gateway.opencode.cloudflare.dev Google inference through pxpipe by @teamchong in https://github.com/teamchong/pxpipe/pull/323 * build: bump actions/setup-node from 7.0.0 to 7.1.0 by @dependabot[bot] in https://github.com/teamchong/pxpipe/pull/317 * build: bump vite from 8.3.0 to 8.3.4 by @dependabot[bot] in https://github.com/teamchong/pxpipe/pull/319 * build: bump @cloudflare/workers-types from 5.20260809.1 to 5.20261011.1 by @dependabot[bot] in https://github.com/teamchong/pxpipe/pull/316 * build: bump @napi-rs/canvas from 1.0.9 to 1.0.11 by @dependabot[bot] in https://github.com/teamchong/pxpipe/pull/315 * build: bump vitest from 4.1.10 to 5.0.4 by @dependabot[bot] in https://github.com/teamchong/pxpipe/pull/310 * build: bump @types/node from 26.2.0 to 26.6.5 by @dependabot[bot] in https://github.com/teamchong/pxpipe/pull/318 * build: bump tsx from 4.23.11 to 4.23.15 by @dependabot[bot] in https://github

  4. v0.14.0v0.14.0Sep 28, 2026

    ## What's Changed * eval(glm): add GLM 5.3 Flash quality benchmarks and native 14px pilot by @teamchong in https://github.com/teamchong/pxpipe/pull/261 * feat(profiles,eval): unify 14px profiles and add 300k/1M LIM reasoning benchmark by @teamchong in https://github.com/teamchong/pxpipe/pull/249 * eval: Fable 5.1 results + Fable 5 control rerun; fix cci.py reply extraction by @teamchong in https://github.com/teamchong/pxpipe/pull/272 * feat(agy, gemini): add agy support and forward-compatible Gemini models by @teamchong in https://github.com/teamchong/pxpipe/pull/271 * Spaced 5x8 profile; enable Opus 5.5 by default by @teamchong in https://github.com/teamchong/pxpipe/pull/293 * build: bump wrangler from 4.120.0 to 4.121.0 by @dependabot[bot] in https://github.com/teamchong/pxpipe/pull/239 * build: bump gpt-tokenizer from 3.4.0 to 4.0.0 by @dependabot[bot] in https://github.com/teamchong/pxpipe/pull/242 * build: bump pnpm/action-setup from 6.0.10 to 6.1.0 by @dependabot[bot] in https://github.com/teamchong/pxpipe/pull/275 * build: bump vite from 8.2.1 to 8.3.0 by @dependabot[bot] in https://github.com/teamchong/pxpipe/pull/283 * build: bump @napi-rs/canvas from 1.0.5 to 1.0.9 by @de

  5. v0.13.2v0.13.2Aug 21, 2026

    ## What's Changed * feat(gemini): add 3.7 Flash with Grok 4.6 and Gemini evals by @teamchong in https://github.com/teamchong/pxpipe/pull/238 * fix(transform): route cc_automode_*/severity/category into dynamic tail by @parziva-1 in https://github.com/teamchong/pxpipe/pull/236 * fix(transform): stop over-truncating reflowed text in truncateForBudget by @dex0shubham in https://github.com/teamchong/pxpipe/pull/226 * perf(render): key the render cache on a digest instead of the source text by @viacheslav-khvorostianyi in https://github.com/teamchong/pxpipe/pull/215 * feat(router): add explicit provider router core by @alteixeira20 in https://github.com/teamchong/pxpipe/pull/223 * build: bump @napi-rs/canvas from 1.0.3 to 1.0.5 by @dependabot[bot] in https://github.com/teamchong/pxpipe/pull/229 * build: bump esbuild from 0.28.1 to 0.28.2 by @dependabot[bot] in https://github.com/teamchong/pxpipe/pull/230 * build: bump @cloudflare/workers-types from 5.20260804.1 to 5.20260809.1 by @dependabot[bot] in https://github.com/teamchong/pxpipe/pull/231 * eval(qwen): add Qwen 3.8 quality benchmarks and native 14px pilot by @teamchong in https://github.com/teamchong/pxpipe/pull/244 * fix(warp): gi

Code frequency

additions and deletions
+10.2M-10.2MWeek of 2026-05-17: +47,380 linesWeek of 2026-05-17: -20,329 linesWeek of 2026-05-24: +2,531 linesWeek of 2026-05-24: -2,212 linesWeek of 2026-05-31: +981 linesWeek of 2026-05-31: -81 linesWeek of 2026-06-07: +52,965 linesWeek of 2026-06-07: -2,848 linesWeek of 2026-06-14: +10,677 linesWeek of 2026-06-14: -4,897 linesWeek of 2026-06-21: +5,638 linesWeek of 2026-06-21: -479 linesWeek of 2026-06-28: +4,080 linesWeek of 2026-06-28: -1,914 linesWeek of 2026-07-05: +34,069 linesWeek of 2026-07-05: -2,997 linesWeek of 2026-07-12: +24,154 linesWeek of 2026-07-12: -1,436 linesWeek of 2026-07-19: +40,898 linesWeek of 2026-07-19: -3,875 linesWeek of 2026-07-26: +10,156,944 linesWeek of 2026-07-26: -210 linesWeek of 2026-08-02: +6,024 linesWeek of 2026-08-02: -1,639 linesWeek of 2026-08-09: +1,937 linesWeek of 2026-08-09: -170,254 linesWeek of 2026-08-16: +21,100 linesWeek of 2026-08-16: -399 linesWeek of 2026-08-23: +158 linesWeek of 2026-08-23: -17 linesWeek of 2026-08-30: +22,396 linesWeek of 2026-08-30: -5,982 linesWeek of 2026-09-06: +13 linesWeek of 2026-09-06: -4 linesWeek of 2026-09-13: +0 linesWeek of 2026-09-13: -0 linesWeek of 2026-09-20: +3,710 linesWeek of 2026-09-20: -308 linesWeek of 2026-09-27: +4,157 linesWeek of 2026-09-27: -54 linesWeek of 2026-10-04: +0 linesWeek of 2026-10-04: -0 linesWeek of 2026-10-11: +12,380 linesWeek of 2026-10-11: -745 linesWeek of 2026-10-18: +9,299 linesWeek of 2026-10-18: -599 linesWeek of 2026-10-25: +2,615 linesWeek of 2026-10-25: -401 linesWeek of 2026-11-01: +196 linesWeek of 2026-11-01: -14 linesWeek of 2026-11-08: +1,057 linesWeek of 2026-11-08: -88 linesWeek of 2026-11-15: +2,084 linesWeek of 2026-11-15: -28 linesMay 17, 2026Nov 15, 2026
+10.5M lines added, -221.8K removed over the last year.

Commits per week

last 52 weeks
1240Week of 2025-11-23: 0 commitsWeek of 2025-11-30: 0 commitsWeek of 2025-12-07: 0 commitsWeek of 2025-12-14: 0 commitsWeek of 2025-12-21: 0 commitsWeek of 2025-12-28: 0 commitsWeek of 2026-01-04: 0 commitsWeek of 2026-01-11: 0 commitsWeek of 2026-01-18: 0 commitsWeek of 2026-01-25: 0 commitsWeek of 2026-02-01: 0 commitsWeek of 2026-02-08: 0 commitsWeek of 2026-02-15: 0 commitsWeek of 2026-02-22: 0 commitsWeek of 2026-03-01: 0 commitsWeek of 2026-03-08: 0 commitsWeek of 2026-03-15: 0 commitsWeek of 2026-03-22: 0 commitsWeek of 2026-03-29: 0 commitsWeek of 2026-04-05: 0 commitsWeek of 2026-04-12: 0 commitsWeek of 2026-04-19: 0 commitsWeek of 2026-04-26: 0 commitsWeek of 2026-05-03: 0 commitsWeek of 2026-05-10: 0 commitsWeek of 2026-05-17: 124 commitsWeek of 2026-05-24: 20 commitsWeek of 2026-05-31: 4 commitsWeek of 2026-06-07: 46 commitsWeek of 2026-06-14: 35 commitsWeek of 2026-06-21: 30 commitsWeek of 2026-06-28: 52 commitsWeek of 2026-07-05: 19 commitsWeek of 2026-07-12: 35 commitsWeek of 2026-07-19: 21 commitsWeek of 2026-07-26: 17 commitsWeek of 2026-08-02: 22 commitsWeek of 2026-08-09: 17 commitsWeek of 2026-08-16: 12 commitsWeek of 2026-08-23: 1 commitsWeek of 2026-08-30: 11 commitsWeek of 2026-09-06: 1 commitsWeek of 2026-09-13: 0 commitsWeek of 2026-09-20: 6 commitsWeek of 2026-09-27: 7 commitsWeek of 2026-10-04: 0 commitsWeek of 2026-10-11: 11 commitsWeek of 2026-10-18: 18 commitsWeek of 2026-10-25: 5 commitsWeek of 2026-11-01: 3 commitsWeek of 2026-11-08: 3 commitsWeek of 2026-11-15: 5 commitsNov 23, 2025Nov 15, 2026
525 commits in the last 52 weeks.

When work happens

weekday and hour
SunMonTueWedThuFriSat036912151821Sun 0:00 — 5 commitsSun 1:00 — 3 commitsSun 2:00 — 1 commitsSun 3:00 — 0 commitsSun 4:00 — 13 commitsSun 5:00 — 5 commitsSun 6:00 — 3 commitsSun 7:00 — 0 commitsSun 8:00 — 1 commitsSun 9:00 — 2 commitsSun 10:00 — 1 commitsSun 11:00 — 2 commitsSun 12:00 — 1 commitsSun 13:00 — 1 commitsSun 14:00 — 2 commitsSun 15:00 — 2 commitsSun 16:00 — 1 commitsSun 17:00 — 8 commitsSun 18:00 — 7 commitsSun 19:00 — 1 commitsSun 20:00 — 6 commitsSun 21:00 — 0 commitsSun 22:00 — 0 commitsSun 23:00 — 0 commitsMon 0:00 — 1 commitsMon 1:00 — 0 commitsMon 2:00 — 0 commitsMon 3:00 — 0 commitsMon 4:00 — 0 commitsMon 5:00 — 1 commitsMon 6:00 — 2 commitsMon 7:00 — 0 commitsMon 8:00 — 2 commitsMon 9:00 — 1 commitsMon 10:00 — 3 commitsMon 11:00 — 5 commitsMon 12:00 — 1 commitsMon 13:00 — 1 commitsMon 14:00 — 3 commitsMon 15:00 — 11 commitsMon 16:00 — 6 commitsMon 17:00 — 10 commitsMon 18:00 — 1 commitsMon 19:00 — 1 commitsMon 20:00 — 0 commitsMon 21:00 — 7 commitsMon 22:00 — 7 commitsMon 23:00 — 4 commitsTue 0:00 — 0 commitsTue 1:00 — 0 commitsTue 2:00 — 0 commitsTue 3:00 — 0 commitsTue 4:00 — 0 commitsTue 5:00 — 0 commitsTue 6:00 — 3 commitsTue 7:00 — 2 commitsTue 8:00 — 4 commitsTue 9:00 — 3 commitsTue 10:00 — 1 commitsTue 11:00 — 6 commitsTue 12:00 — 6 commitsTue 13:00 — 5 commitsTue 14:00 — 7 commitsTue 15:00 — 5 commitsTue 16:00 — 4 commitsTue 17:00 — 5 commitsTue 18:00 — 7 commitsTue 19:00 — 10 commitsTue 20:00 — 3 commitsTue 21:00 — 11 commitsTue 22:00 — 5 commitsTue 23:00 — 3 commitsWed 0:00 — 3 commitsWed 1:00 — 0 commitsWed 2:00 — 1 commitsWed 3:00 — 0 commitsWed 4:00 — 0 commitsWed 5:00 — 1 commitsWed 6:00 — 2 commitsWed 7:00 — 7 commitsWed 8:00 — 1 commitsWed 9:00 — 10 commitsWed 10:00 — 12 commitsWed 11:00 — 2 commitsWed 12:00 — 3 commitsWed 13:00 — 5 commitsWed 14:00 — 1 commitsWed 15:00 — 4 commitsWed 16:00 — 5 commitsWed 17:00 — 1 commitsWed 18:00 — 4 commitsWed 19:00 — 4 commitsWed 20:00 — 3 commitsWed 21:00 — 2 commitsWed 22:00 — 3 commitsWed 23:00 — 6 commitsThu 0:00 — 0 commitsThu 1:00 — 0 commitsThu 2:00 — 0 commitsThu 3:00 — 0 commitsThu 4:00 — 1 commitsThu 5:00 — 1 commitsThu 6:00 — 0 commitsThu 7:00 — 2 commitsThu 8:00 — 3 commitsThu 9:00 — 0 commitsThu 10:00 — 5 commitsThu 11:00 — 3 commitsThu 12:00 — 9 commitsThu 13:00 — 3 commitsThu 14:00 — 1 commitsThu 15:00 — 4 commitsThu 16:00 — 3 commitsThu 17:00 — 12 commitsThu 18:00 — 1 commitsThu 19:00 — 3 commitsThu 20:00 — 0 commitsThu 21:00 — 5 commitsThu 22:00 — 9 commitsThu 23:00 — 8 commitsFri 0:00 — 11 commitsFri 1:00 — 2 commitsFri 2:00 — 0 commitsFri 3:00 — 0 commitsFri 4:00 — 0 commitsFri 5:00 — 2 commitsFri 6:00 — 3 commitsFri 7:00 — 0 commitsFri 8:00 — 2 commitsFri 9:00 — 4 commitsFri 10:00 — 0 commitsFri 11:00 — 4 commitsFri 12:00 — 3 commitsFri 13:00 — 4 commitsFri 14:00 — 1 commitsFri 15:00 — 1 commitsFri 16:00 — 6 commitsFri 17:00 — 0 commitsFri 18:00 — 6 commitsFri 19:00 — 2 commitsFri 20:00 — 3 commitsFri 21:00 — 1 commitsFri 22:00 — 14 commitsFri 23:00 — 4 commitsSat 0:00 — 0 commitsSat 1:00 — 0 commitsSat 2:00 — 0 commitsSat 3:00 — 0 commitsSat 4:00 — 2 commitsSat 5:00 — 1 commitsSat 6:00 — 0 commitsSat 7:00 — 1 commitsSat 8:00 — 0 commitsSat 9:00 — 1 commitsSat 10:00 — 1 commitsSat 11:00 — 2 commitsSat 12:00 — 10 commitsSat 13:00 — 4 commitsSat 14:00 — 4 commitsSat 15:00 — 1 commitsSat 16:00 — 3 commitsSat 17:00 — 0 commitsSat 18:00 — 8 commitsSat 19:00 — 0 commitsSat 20:00 — 13 commitsSat 21:00 — 13 commitsSat 22:00 — 6 commitsSat 23:00 — 7 commits
Commit volume by weekday and hour (UTC). Larger dots mean more commits.

Who is committing

last 52 weeks
Maintainer commits445 (82%)
Community commits98 (18%)

543 commits in total over the last year.

DateListRankStars gained
Jul 6, 2026daily#6+8
Jul 5, 2026daily#14+24
Jul 4, 2026daily#9+33
  • morluto/rea

    Reverse engineer anything with agents, from app behavior down to native binaries.

    471.6K stars · TypeScript

  • freeCodeCamp/freeCodeCamp

    freeCodeCamp.org's open-source codebase and curriculum. Learn math, programming, and computer science for free.

    459.4K stars · TypeScript

  • openclaw/openclaw

    The AI that really does things. Any OS. Any Platform. The lobster way. 🦞

    394.9K stars · TypeScript

  • anomalyco/opencode

    The open source coding agent.

    228.3K stars · TypeScript

  • n8n-io/n8n

    Fair-code workflow automation platform with native AI capabilities. Combine visual building with custom code, self-host or cloud, 400+ integrations.

    211.6K stars · TypeScript

  • firecrawl/firecrawl

    Knowledge infrastructure to make your agents smarter. Search, scrape and access 550+ data providers through one API.

    205.3K stars · TypeScript