Claude Fable 5.1 vs GPT-5.6 Sol: Benchmark Comparison
Claude Fable 5.1 has the higher score on 27 of 32 shared benchmarks; GPT-5.6 Sol leads on 5.
The largest observed score gap is 39.50 pts on Time Horizon Index: KSP , where Claude Fable 5.1 leads.
Reported ±1 standard-error ranges overlap on 9 of 30 shared benchmarks with comparable uncertainty data. This is not a pairwise statistical significance test.
Shared benchmark results
Scores and reported standard errors come from the latest visible Vals benchmark version available for both models.
| Benchmark | Claude Fable 5.1 | GPT-5.6 Sol | Gap | Reported uncertainty |
|---|---|---|---|---|
| Vals Index | 65.83% ±1.11 | 58.01% ±1.02 | 7.82 pts | Reported ±1 SE ranges do not overlap |
| Vals RSI Index | 36.09% | 23.88% | 12.21 pts | Uncertainty comparison unavailable |
| Harvey's Legal Agent Benchmark | 6.67% ±1.17 | 2.50% ±0.83 | 4.17 pts | Reported ±1 SE ranges do not overlap |
| Legal Research Bench | 55.29% ±3.46 | 48.08% ±3.47 | 7.21 pts | Reported ±1 SE ranges do not overlap |
| LegalBench | 88.51% ±0.42 | 86.97% ±0.41 | 1.55 pts | Reported ±1 SE ranges do not overlap |
| EMB | 76.67% ±2.08 | 72.34% ±2.37 | 4.33 pts | Reported ±1 SE ranges overlap |
| Finance Agent (v2) | 58.88% ±2.06 | 53.76% ±0.85 | 5.12 pts | Reported ±1 SE ranges do not overlap |
| MortgageTax | 70.79% ±0.90 | 67.29% ±0.92 | 3.50 pts | Reported ±1 SE ranges do not overlap |
| Tax Agent Bench | 77.64% ±2.83 | 67.95% ±3.12 | 9.69 pts | Reported ±1 SE ranges do not overlap |
| TaxEval v2 | 75.96% ±0.83 | 74.78% ±0.86 | 1.19 pts | Reported ±1 SE ranges overlap |
| MedCode | 53.51% ±2.17 | 43.97% ±2.26 | 9.54 pts | Reported ±1 SE ranges do not overlap |
| MedScribe | 91.29% ±1.95 | 85.23% ±1.97 | 6.06 pts | Reported ±1 SE ranges do not overlap |
| ProofBench v1.1 | 100.00% ±0.00 | 83.00% ±3.77 | 17.00 pts | Reported ±1 SE ranges do not overlap |
| MysteryMechanism | 47.75% ±3.36 | 33.33% ±3.17 | 14.41 pts | Reported ±1 SE ranges do not overlap |
| Terminal-Bench Science | 40.00% ±5.90 | 20.00% ±4.82 | 20.00 pts | Reported ±1 SE ranges do not overlap |
| GPQA Diamond | 93.43% ±1.86 | 95.20% ±1.07 | 1.77 pts | Reported ±1 SE ranges overlap |
| MMLU Pro | 92.38% ±0.27 | 89.10% ±0.31 | 3.28 pts | Reported ±1 SE ranges do not overlap |
| MMMU Pro | 90.64% ±0.70 | 88.84% ±0.76 | 1.79 pts | Reported ±1 SE ranges do not overlap |
| SAGE | 48.53% ±3.34 | 52.56% ±3.42 | 4.04 pts | Reported ±1 SE ranges overlap |
| Code Migration | 54.61% ±4.81 | 52.92% ±4.35 | 1.69 pts | Reported ±1 SE ranges overlap |
| IOI | 90.78% ±4.65 | 91.17% ±4.51 | 0.39 pts | Reported ±1 SE ranges overlap |
| LiveCodeBench | 90.52% ±0.86 | 82.60% ±1.09 | 7.91 pts | Reported ±1 SE ranges do not overlap |
| ProgramBench | 7.00% ±1.81 | 1.50% ±0.86 | 5.50 pts | Reported ±1 SE ranges do not overlap |
| SkillsBench | 61.55% ±4.76 | 54.10% ±4.73 | 7.45 pts | Reported ±1 SE ranges overlap |
| Terminal-Bench 4.0 | 58.08% ±3.31 | 37.88% ±0.00 | 20.20 pts | Reported ±1 SE ranges do not overlap |
| Vibe Code Bench 1-100 | 28.00% ±4.49 | 20.01% ±3.63 | 7.99 pts | Reported ±1 SE ranges overlap |
| Vibe Code Bench v1.1 | 90.26% ±1.57 | 80.50% ±3.72 | 9.77 pts | Reported ±1 SE ranges do not overlap |
| CyberBench v1.1 | 70.42% ±5.43 | 76.31% ±5.18 | 5.89 pts | Reported ±1 SE ranges overlap |
| SRE Bench | 22.90% ±2.60 | 30.53% ±2.85 | 7.63 pts | Reported ±1 SE ranges do not overlap |
| Public Benefits Bench v1.1 | 74.90% ±1.13 | 66.51% ±1.23 | 8.39 pts | Reported ±1 SE ranges do not overlap |
| CUA-bench | 13.17% | 8.33% | 4.83 pts | Uncertainty comparison unavailable |
| Time Horizon Index: KSP | 63.33% ±0.00 | 23.83% ±0.00 | 39.50 pts | Reported ±1 SE ranges do not overlap |
Performance by category
| Category | Claude Fable 5.1 average | GPT-5.6 Sol average |
|---|---|---|
| Legal | 50.16% | 45.85% |
| Finance | 71.99% | 67.22% |
| Healthcare | 72.40% | 64.60% |
| Math | 100.00% | 83.00% |
| Science | 43.87% | 26.67% |
| Academic | 92.15% | 91.05% |
| Education | 48.53% | 52.56% |
| Coding | 60.10% | 52.58% |
| Cyber | 46.66% | 53.42% |
| Social Mobility | 74.90% | 66.51% |
Cost and latency
Cost per test appears only where the benchmark reports it for both models. Latency is the measured completion time for that benchmark.
| Benchmark | Claude Fable 5.1 cost | GPT-5.6 Sol cost | Claude Fable 5.1 latency | GPT-5.6 Sol latency |
|---|---|---|---|---|
| Vals Index | $28.71 | $14.24 | 1h17m | 37m23s |
| Vals RSI Index | $298.22 | $347.10 | 90h00m | 90h00m |
| Harvey's Legal Agent Benchmark | $46.21 | $10.37 | 1h52m | 22m50s |
| Legal Research Bench | $23.06 | $21.61 | 57m29s | 1h17m |
| LegalBench | N/A | N/A | 12.40s | 6.20s |
| EMB | $15.93 | $6.01 | 29m32s | 9m50s |
| Finance Agent (v2) | $8.35 | $1.25 | 18m14s | 19m25s |
| MortgageTax | N/A | N/A | 14.98s | 12.62s |
| Tax Agent Bench | $13.18 | $7.51 | 28m34s | 44m41s |
| TaxEval v2 | N/A | N/A | 85.56s | 62.03s |
| MedCode | N/A | N/A | 3m35s | 96.59s |
| MedScribe | N/A | N/A | 3m08s | 94.40s |
| ProofBench v1.1 | $2.70 | $1.44 | 7m58s | 8m40s |
| MysteryMechanism | $5.63 | $0.96 | 17m41s | 8m43s |
| Terminal-Bench Science | $38.01 | $7.32 | 2h36m | 1h44m |
| GPQA Diamond | N/A | N/A | 45.86s | 58.30s |
| MMLU Pro | N/A | N/A | 20.59s | 15.65s |
| MMMU Pro | N/A | N/A | 37.47s | 26.20s |
| SAGE | N/A | N/A | 105.75s | 75.39s |
| Code Migration | $70.97 | $24.54 | 4h20m | 59m12s |
| IOI | $11.20 | $7.98 | 30m06s | 1h02m |
| LiveCodeBench | N/A | N/A | 59.74s | 56.51s |
| ProgramBench | $58.52 | $15.29 | 2h27m | 29m24s |
| SkillsBench | $4.52 | $4.14 | 11m02s | 13m07s |
| Terminal-Bench 4.0 | $17.18 | $7.98 | 1h00m | 33m46s |
| Vibe Code Bench 1-100 | $149.66 | $47.43 | 2h53m | 2h37m |
| Vibe Code Bench v1.1 | $33.37 | $33.40 | 57m40s | 33m53s |
| CyberBench v1.1 | $4.01 | $3.05 | 18m32s | 10m47s |
| SRE Bench | $32.23 | $42.52 | 2h03m | 1h28m |
| Public Benefits Bench v1.1 | $7.48 | $8.85 | 48m02s | 1h01m |
| CUA-bench | $323.78 | $206.99 | 37.44s | 18.80s |
| Time Horizon Index: KSP | $4655.01 | $2027.20 | N/A | N/A |
Results available only for Claude Fable 5.1
None.
Results available only for GPT-5.6 Sol
- BioMysteryBench
- SWE-bench