Claude Haiku 5.5 vs DeepSeek V4.1 Flash 1017: Benchmark Comparison
Claude Haiku 5.5 has the higher score on 3 of 9 shared benchmarks; DeepSeek V4.1 Flash 1017 leads on 6.
The largest observed score gap is 12.43 pts on Tax Agent Bench , where DeepSeek V4.1 Flash 1017 leads.
Reported ±1 standard-error ranges overlap on 7 of 9 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 Haiku 5.5 | DeepSeek V4.1 Flash 1017 | Gap | Reported uncertainty |
|---|---|---|---|---|
| Vals Index | 54.44% ±1.41 | 56.58% ±1.27 | 2.15 pts | Reported ±1 SE ranges overlap |
| Harvey's Legal Agent Benchmark | 1.25% ±0.72 | 9.17% ±2.51 | 7.92 pts | Reported ±1 SE ranges do not overlap |
| Legal Research Bench | 43.27% ±3.44 | 45.19% ±3.46 | 1.92 pts | Reported ±1 SE ranges overlap |
| EMB | 59.02% ±4.17 | 61.80% ±3.19 | 2.79 pts | Reported ±1 SE ranges overlap |
| Finance Agent (v2) | 54.14% ±2.13 | 56.96% ±2.05 | 2.82 pts | Reported ±1 SE ranges overlap |
| Tax Agent Bench | 31.92% ±3.35 | 44.35% ±3.57 | 12.43 pts | Reported ±1 SE ranges do not overlap |
| Code Migration | 44.57% ±4.23 | 44.21% ±4.46 | 0.36 pts | Reported ±1 SE ranges overlap |
| Terminal-Bench 4.0 | 36.36% ±2.31 | 31.82% ±2.62 | 4.55 pts | Reported ±1 SE ranges overlap |
| Vibe Code Bench v1.1 | 90.44% ±1.73 | 87.73% ±2.75 | 2.71 pts | Reported ±1 SE ranges overlap |
Performance by category
| Category | Claude Haiku 5.5 average | DeepSeek V4.1 Flash 1017 average |
|---|---|---|
| Legal | 22.26% | 27.18% |
| Finance | 48.36% | 54.37% |
| Coding | 57.13% | 54.59% |
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 Haiku 5.5 cost | DeepSeek V4.1 Flash 1017 cost | Claude Haiku 5.5 latency | DeepSeek V4.1 Flash 1017 latency |
|---|---|---|---|---|
| Vals Index | $3.00 | $0.64 | 47m24s | 36m33s |
| Harvey's Legal Agent Benchmark | $3.39 | $0.54 | 35m14s | 21m39s |
| Legal Research Bench | $1.63 | $0.70 | 1h22m | 35m07s |
| EMB | $1.15 | $0.36 | 18m03s | 16m47s |
| Finance Agent (v2) | $1.64 | $0.46 | 30m49s | 14m39s |
| Tax Agent Bench | $1.16 | $0.46 | 49m34s | 21m04s |
| Code Migration | $6.23 | $1.86 | 1h23m | 2h11m |
| Terminal-Bench 4.0 | $3.12 | $0.54 | 51m56s | 41m10s |
| Vibe Code Bench v1.1 | $6.07 | $0.58 | 34m36s | 16m14s |
Results available only for Claude Haiku 5.5
- Vals Science Index
- MedCode
- MedScribe
- ProofBench v1.1
- BioMysteryBench
- MysteryMechanism
- Terminal-Bench-Science
- SAGE
- IOI
- ProgramBench
- SWE-bench Multimodal
- Vibe Code Bench 1-100
- CyberBench v1.1
- SRE Bench
- Public Benefits Bench v1.1
Results available only for DeepSeek V4.1 Flash 1017
None.