Agent
Claude Code
Claude Code is an AI coding agent made by Anthropic.
- Episodes
- 0
- Published runs
- 20
What it built · Lava lamp v1
4 models, 20 runs.
| Model | Low4 models | Medium4 models | High4 models | Extra high4 models | Max4 models |
|---|---|---|---|---|---|
| Claude Code | |||||
| Claude Fable 5.15 runs · 5 of 5 work | |||||
| Claude Haiku 5.55 runs · 5 of 5 work | |||||
| Claude Opus 5.55 runs · 5 of 5 work | |||||
| Claude Sonnet 5.55 runs · 5 of 5 work | |||||
Low effort · 4 models · fastest first
Medium effort · 4 models · fastest first
High effort · 4 models · fastest first
Extra high effort · 4 models · fastest first
Max effort · 4 models · fastest first
Build time: the number is the median, each dot one run, on one scale per effort.
About Claude Code
About
- Vendor
- Anthropic
- Official site
- https://claude.com/product/claude-code
- CmdBench support
- runner
What this repository records
Only facts a file in this repository proves, with that file. Each run records its own harness version, binary hash, model, effort and plan; nothing on this page overrides a run.
- CmdBench runner support exists (provider "claude", harness label "Claude CLI").Source: src/server.ts
- The CLI binary is resolved on PATH and its SHA-256 is recorded per run.Source: src/runner/binary.ts
- Models used by existing presets are claude-sonnet-5-5 (runner default) and claude-opus-5-5.Source: src/runner/settings.ts, src/server.ts
- Default generation deadline is 600000 ms.Source: src/runner/settings.ts
- Effort is recorded as awaiting OTLP request spans until telemetry reports it.Source: src/runner/index.ts
Not recorded yet
- pinned harness version for cb-001 (spec readiness gate asks for >= 2.1.304; not yet recorded)
- binary SHA-256 for cb-001
- native image attachment support and input transport for v2 benchmarks
- plan used for runs (subscription or API)
- usage fields reported (tokens, cost, plan usage)
Versions and models tested
| Harness version | Model | Effort | Runs | First run | Last run |
|---|---|---|---|---|---|
| 2.1.312 (Claude Code) | claude-opus-5-5 | max | 1 | 28.10.2026 14:40 | 28.10.2026 14:40 |
| 2.1.312 (Claude Code) | claude-sonnet-5-5 | max | 1 | 28.10.2026 13:12 | 28.10.2026 13:12 |
| 2.1.312 (Claude Code) | claude-fable-5-1 | max | 1 | 28.10.2026 12:09 | 28.10.2026 12:09 |
| 2.1.312 (Claude Code) | claude-fable-5-1 | xhigh | 1 | 28.10.2026 11:35 | 28.10.2026 11:35 |
| 2.1.312 (Claude Code) | claude-fable-5-1 | high | 1 | 28.10.2026 11:20 | 28.10.2026 11:20 |
| 2.1.312 (Claude Code) | claude-fable-5-1 | medium | 1 | 28.10.2026 11:14 | 28.10.2026 11:14 |
| 2.1.312 (Claude Code) | claude-fable-5-1 | low | 1 | 28.10.2026 11:10 | 28.10.2026 11:10 |
| 2.1.312 (Claude Code) | claude-opus-5-5 | xhigh | 1 | 28.10.2026 09:39 | 28.10.2026 09:39 |
| 2.1.312 (Claude Code) | claude-opus-5-5 | high | 1 | 28.10.2026 09:25 | 28.10.2026 09:25 |
| 2.1.312 (Claude Code) | claude-sonnet-5-5 | xhigh | 1 | 28.10.2026 08:01 | 28.10.2026 08:01 |
| 2.1.312 (Claude Code) | claude-sonnet-5-5 | high | 1 | 28.10.2026 07:54 | 28.10.2026 07:54 |
| 2.1.312 (Claude Code) | claude-haiku-5-5 | medium | 1 | 28.10.2026 07:47 | 28.10.2026 07:47 |
| 2.1.312 (Claude Code) | claude-haiku-5-5 | low | 1 | 28.10.2026 07:44 | 28.10.2026 07:44 |
| 2.1.293 (Claude Code) | claude-haiku-5-5 | max | 1 | 07.10.2026 21:39 | 07.10.2026 21:39 |
| 2.1.293 (Claude Code) | claude-haiku-5-5 | xhigh | 1 | 07.10.2026 21:28 | 07.10.2026 21:28 |
| 2.1.293 (Claude Code) | claude-haiku-5-5 | high | 1 | 07.10.2026 21:22 | 07.10.2026 21:22 |
| 2.1.285 (Claude Code) | claude-opus-5-5 | low | 1 | 30.09.2026 19:51 | 30.09.2026 19:51 |
| 2.1.285 (Claude Code) | claude-opus-5-5 | medium | 1 | 30.09.2026 19:41 | 30.09.2026 19:41 |
| 2.1.285 (Claude Code) | claude-sonnet-5-5 | low | 1 | 30.09.2026 16:08 | 30.09.2026 16:08 |
| 2.1.285 (Claude Code) | claude-sonnet-5-5 | medium | 1 | 30.09.2026 16:03 | 30.09.2026 16:03 |