Published reference2026-06-09
Terminal-Bench
Version 2.1 · sample size not reported. Compare only matched task sets and protocols.
Claude Fable 5 · Anthropic system card · Anthropic · Anthropic evaluation
↗ Published resultProtocol as described by the publisher
Claude Fable 5 and Mythos 5 System Card, sections 8.2–8.3 ↗Tokens per second · see measurement basis
As recorded on this receipt
Run configuration & provenance +
- Context
- Not recorded
- Temperature
- Not recorded
- Wall time
- Not recorded
- Weight file
- Provider-managed / not reported
- Revision
- Not reported
- Reasoning
- Not reported
- Agent version
- Not reported
- Output token limit
- Not reported
- Dataset revision
- Not reported
- Harness host
- Not reported
- Model host
- Published reference
- Input tokens
- Not reported
- Output tokens
- Not reported
- Cached input tokens
- Not reported
- Agent turn / step limit
- Not reported
- Concurrency
- Not reported
- Weight file size
- Not reported
Throughput measurement: Measurement method not separately reported. This should not be treated as standardized decode-only speed.
Provider evaluation
Section 8.3: high effort, mini-SWE-agent on GKE, 1x timeout and 3x memory ceiling. 20.9% of trials fell back to Opus 4.8 after a safety refusal. Mean reward; do not treat as Claude Code max. GPQA in section 8.8 is reported for Mythos 5 only; no Fable GPQA value is inferred.
Run ID: research-claude-fable-5-claude-code-max-anthropic-system-card-terminal-bench-2.1Task explorer
Passed tasks first, then other outcomes grouped by recorded issue. Use the filters to inspect a particular outcome.
No task records in this receipt.
The headline or summary is available above. Individual outcomes have not been provided.
Explore test coverage →