gpt-load
Self-hosted AI gateway for multi-channel, multi-credential setups, API keys and subscription accounts, scheduling, failover, request logs and usage. 自托管 AI 网关：多渠道多凭据统一接入，含密钥与订阅账号、调度容错、日志与用量。
Runssource: GitHubhomepageGoMITcommit ae59a0ae8515
Go, MIT licensed. The project labels itself: ai gateway, anthropic, api gateway, claude, claude code, codex, gemini and gin.
gpt-load runs. An Argusic agent installed it in 45 minutes and hit 5 errors and fixed every one without help on a clean machine with no GPU, and the whole session was recorded.
Measured by Argusic on a fresh machine every time. Every number links to its evidence. Argusic Score 100 of 100 (the mean of this project's run scores).
At a glance
Subject data from GitHub, linked at the top of this page, refreshed . Test data by Argusic (CC BY 4.0); every number links to a run page with the full log, the recording, and their sha256 hashes.
What was measured
From the run that decided this verdict. Every line is a recorded value, not an estimate.
What went wrong, in the agent's own words
The agent's own words, worst first, never rewritten into advice.
- Go compiler not found in environment3 minutes
- Node.js v18.19.1 too old; web UI requires >=24.11.02 minutes
- corepack not in PATH so pnpm was unavailable2 minutes
- Web UI build failed: @rolldown/binding-linux-x64-gnu missing (frozen lockfile skips optional deps)2 minutes
- Docker compose contract tests fail: docker not in PATH (6 subtest failures)
- Install time
- 45 minutes
- Cold machine to finish
- 20 minutes
- Errors hit and fixed
- 5 hit, 5 fixed with no human help
- How the result was proved
- curl http://127.0.0.1:3001/health returned 200 with {"status":"ok","version":"2.0.0-dev"}; curl http://127.0.0.1:3001/ returned 200 with valid HTML; go test -count=1 . ./internal/... exited 1 only for Docker-dependent tests (58/58 Go packages passed)
- Model tokens used
- 126,517
- Exact commit tested
- ae59a0ae8515
- Container image
- sha256:cdd920bce783
What was not measured
The other half, so you can see exactly how far this evidence goes. A gap is printed as a gap and never filled in.
- Peak GPU memory
- this ran on a CPU machine, so there was no GPU to sample
- Behaviour over a long run
- each attempt uses the smallest documented settings and a fixed time budget, so this measures whether it runs, not how it holds up under load
Watch the session
The whole run, from a clean machine, at the speed it actually happened. Nothing is cut and nothing is sped up.
A recording of attempt 2, run , the attempt that decided the verdict above. The terminal recording it was made from is the evidence, stored with a sha256 fingerprint; the video is a rendering of it. Nothing loads from the video host until you press play.
The end of the recording
The last lines of the terminal session that decided this verdict, exactly as they were recorded.
time="2026-09-16 16:36:00" level=info msg="committed control operations recovered" event=startup.op…
time="2026-09-16 16:36:00" level=info msg="runtime state checkpoint checked" event=startup.checkpoi…
time="2026-09-16 16:36:00" level=info msg="execution runtime started" event=startup.execution_runti…
time="2026-09-16 16:36:00" level=info msg="HTTP listener bound" event=startup.server_listen address…
time="2026-09-16 16:36:00" level=info msg="request log runtime started" event=startup.request_log_s…
time="2026-09-16 16:36:00" level=info msg="GPT-Load 2.0 server started" event=startup.ready address…
time="2026-09-16 16:36:00" level=info msg="control runtime started" event=startup.control_runtime_s…
time="2026-09-16 16:36:00" level=info msg="[CONTROL] Models.dev catalog synchronization started" ev…
200
{"status":"ok","version":"2.0.0-dev"}
tokens used
126,517Replay the whole session, every command from a clean machine to this point.
How it was tested
One agent, one container, no human help. It installs the project the way its own documentation says to, runs it, and fixes what breaks. Everything below is recorded as it happened: the terminal session, the log and the exact commit. The full method.
Strengths and limits
Measured facts, not opinions. How this is written.
What went well
- Reached a running state on a clean machine, with the session recorded.
- Installed in 45 minutes.
- Recovered from all 5 errors without a human stepping in, which says the failures are documented well enough to solve.
- Ran without a GPU, so it does not need one to start.
- Licensed MIT, as reported by its host.
- Measured 2 times, so the result is not a one-off.
What it is a reasonable choice for
- Trying it on a laptop or a small server: it reached a running state without a GPU.
Also tested, in the same area
Every one of these was installed and run by Argusic on a clean machine. Nothing appears here that was not tested.
Run history
Topics (from GitHub)
ai-gatewayanthropicapi-gatewayclaudeclaude-codecodexgeminigingogolangllmllm-gatewayload-balanceropenaiself-hosted
Embed the badge
Markdown for the project README. It links back here; terms on the terms page.
[](https://argusic.com/subject/gpt-load)Questions
- Does gpt-load run?
- Yes. gpt-load runs. Argusic installed and launched it on a clean machine in 45 minutes, hitting 5 errors on the way, and recorded the session.
- How did Argusic test gpt-load?
- On a fresh, disposable machine, with every command recorded and the repository pinned at commit ae59a0ae8515. 2 attempts are recorded, and the full method is on the methodology page.
- What did it cost to test gpt-load?
- The run that produced this verdict cost $0.09: real compute and model cost, not a list price per million tokens. It is recorded on that run's page.
- How long does gpt-load take to install?
- 45 minutes in the run that decided this verdict, from a clean machine to a running state. That is one measured attempt, not an average.
- Does gpt-load need a GPU?
- Not to start. Argusic reached a running state on a machine with no GPU.
- What goes wrong when installing gpt-load?
- 5 things broke in the recorded run, and 5 were fixed without human help. Each one, and the time it cost, is listed on this page.
- Where is the evidence for gpt-load?
- All 2 recorded runs are on this page, each linking to its full log and terminal recording, stored with a sha256 fingerprint so it cannot be quietly altered.