GPT-5.6 SolvsClaude Fable 5
Across 14 shared benchmarks, Claude Fable 5 leads overall: GPT-5.6 Sol wins 5, Claude Fable 5 wins 9, with 0 ties and an average score difference of -8.51.
GPT-5.6 Sol
OpenAI · 2026-06-26 · Reasoning model
Claude Fable 5
Anthropic · 2026-06-09 · Reasoning model
GPT-5.6 Sol5 wins(36%)(64%)9 winsClaude Fable 5
Benchmark scores
Grouped by capability, sorted by largest gap within each. 14 shared benchmarks.
Coding and Software Engineer
Claude Fable 5 3/5| Benchmark | GPT-5.6 Sol | Claude Fable 5 | Diff |
|---|---|---|---|
| SWE-Bench Pro - Public | 64.607 / 57极高强度思考（工具） | 80.301 / 57Deep Thinking (With Tools) | -15.70 |
| CursorBench 3.2 | 67.203 / 4Max (With Tools) | 70.501 / 4Max (With Tools) | -3.30 |
| FrontierCode 1.1 | 60.603 / 5Max (With Tools) | 63.601 / 5Max (With Tools) | -3 |
| DeepSWE | 72.701 / 27极高强度思考（工具） | 702 / 27Deep Thinking (With Tools) | +2.70 |
| WeirdML v2 | 88.762 / 52Thinking High (With Tools) | 87.853 / 52Thinking High (With Tools) | +0.91 |
Productivity Knowledge
Claude Fable 5 3/3| Benchmark | GPT-5.6 Sol | Claude Fable 5 | Diff |
|---|---|---|---|
| AA-Briefcase | 1,5025 / 6Max (With Tools) | 1,5743 / 6Max (With Tools) | -72 |
| GDPval-AA v2 | 1,7285 / 13Max (With Tools) | 1,7414 / 13Max (With Tools) | -13 |
| Harvey Lab-AA | 2.506 / 6Max (With Tools) | 11.305 / 6Max (With Tools) | -8.80 |
AI Agent - Tool Usage
GPT-5.6 Sol 2/2| Benchmark | GPT-5.6 Sol | Claude Fable 5 | Diff |
|---|---|---|---|
| Terminal-Bench 2.1 | 88.801 / 44最高（无工具） | 884 / 44Deep Thinking (With Tools) | +0.80 |
| Terminal-Bench 3.0 | 34.601 / 6Max (With Tools) | 34.102 / 6Max (With Tools) | +0.50 |
Math and Reasoning
Even 2/2| Benchmark | GPT-5.6 Sol | Claude Fable 5 | Diff |
|---|---|---|---|
| FrontierMath Tier 4 v2 | 82.932 / 34最高（无工具） | 87.801 / 34最高（无工具） | -4.88 |
| FrontierMath v2 | 89.121 / 34最高（无工具） | 87.023 / 34最高（无工具） | +2.11 |
Agent Level Benchmark
Claude Fable 5 1/1| Benchmark | GPT-5.6 Sol | Claude Fable 5 | Diff |
|---|---|---|---|
| APEX-Agents | 56.703 / 5Max (With Tools) | 59.201 / 5Max (With Tools) | -2.50 |
General Knowledge
Claude Fable 5 1/1| Benchmark | GPT-5.6 Sol | Claude Fable 5 | Diff |
|---|---|---|---|
| AA Intelligence Index | 593 / 8最高（无工具） | 621 / 8Max (With Tools) | -3 |
Specs
| Field | GPT-5.6 Sol | Claude Fable 5 |
|---|---|---|
| Publisher | OpenAI | Anthropic |
| Release date | 2026-06-26 | 2026-06-09 |
| Model type | Reasoning model | Reasoning model |
| Architecture | Dense | Dense |
| Parameters | Not available | Not available |
| Context length | 1.05M | 1M |
| Max output | 128K | 128K |
API pricing
Prices use DataLearner records when available; missing fields are not inferred.
| Item | GPT-5.6 Sol | Claude Fable 5 |
|---|---|---|
| Text input | $5 / 1M tokens | $10 / 1M tokens |
| Text output | $30 / 1M tokens | $50 / 1M tokens |
| Cache read | $0.5 / 1M tokens | $1 / 1M tokens |
| Cache write | $6.25 / 1M tokens | $12.5 / 1M tokens |
Summary
- GPT-5.6 Solleads in:AI Agent - Tool Usage (2/2)
- Claude Fable 5leads in:Coding and Software Engineer (3/5), Productivity Knowledge (3/3), Agent Level Benchmark (1/1), General Knowledge (1/1)
- Tied in:Math and Reasoning
On average across the 14 shared benchmarks, Claude Fable 5 scores 8.51 higher.
Largest single-benchmark gap: AA-Briefcase — GPT-5.6 Sol 1,502 vs Claude Fable 5 1,574 (-72).
Page generated from structured model, pricing and benchmark records. No real-time LLM is used to write the prose.