Model comparison
Claude Haiku 4.5 vs DeepSeek-R1
DeepSeek-R1 is the stronger model overall, scoring 42.3 to 39.5 on the Noometry Index.
Last verified . 36 shared benchmarks.
DeepSeekSummary
- They share 36 benchmarks with published results for both. Claude Haiku 4.5 scores higher in 2 categories and DeepSeek-R1 in 7 categories; 8 gaps are clear of the uncertainty.
- The widest gap is in knowledge, where DeepSeek-R1 leads 44.5 to 37.7.
- The biggest single-benchmark swing is ARC-AGI-1: 47.7% for Claude Haiku 4.5 and 21.2% for DeepSeek-R1.
- DeepSeek-R1 is cheaper at $0.50 / $2.15 per million input/output tokens, against $1 / $5 for Claude Haiku 4.5.
- Claude Haiku 4.5 accepts more context: 200K tokens versus 164K.
Side by side
| Claude Haiku 4.5 | DeepSeek-R1 | |
|---|---|---|
| Provider | Anthropic | DeepSeek |
| Noometry Index | 39.5 | 42.3 |
| Released | 2025-10-15 | 2025-01-20 |
| Weights | Proprietary | Proprietary |
| Context window | 200K | 164K |
| Max output | 64K | 64K |
| Input $ / M tokens | $1 | $0.50 |
| Output $ / M tokens | $5 | $2.15 |
| Results tracked | 53 | 52 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding DeepSeek-R1 leads
Claude Haiku 4.5: 44.0 (#78), DeepSeek-R1: 46.3 (#68)
| Benchmark | Claude Haiku 4.5 | DeepSeek-R1 |
|---|---|---|
| SciCode | 43.3% | 35.7% |
| WeirdML | 45.4% | 41.6% |
| LMArena Coding | 1453 | 1427 |
| ALE-Bench | 653.48 | 804.12 |
| SWE-bench Verified (bash only) | 66.6% | — |
| Aider Polyglot | — | 71.4% |
| LMArena WebDev | 1330 | — |
| SWE-bench Multilingual | 64.7% | — |
| LiveBench Coding | — | 66.7% |
| AlgoTune | — | 1.7 |
Agentic & Tool Use Claude Haiku 4.5 leads
Claude Haiku 4.5: 33.6 (#52), DeepSeek-R1: 30.7 (#75)
| Benchmark | Claude Haiku 4.5 | DeepSeek-R1 |
|---|---|---|
| DeepResearch Bench | 45.5% | 35.1% |
| BALROG | 31.2% | 34.9% |
| Terminal-Bench | 35.5% | — |
| Berkeley Function Calling Leaderboard | 68.7% | — |
| ExploitBench | 13.7% | — |
| METR Time Horizons | — | 53.8% |
| Vending-Bench 2 | 458.89 | — |
Reasoning DeepSeek-R1 leads
Claude Haiku 4.5: 15.1 (#320), DeepSeek-R1: 18.6 (#278)
| Benchmark | Claude Haiku 4.5 | DeepSeek-R1 |
|---|---|---|
| ARC-AGI-2 | 4% | 1.3% |
| ARC-AGI-1 | 47.7% | 21.2% |
| CritPt | 0% | 1.1% |
| LMArena Hard Prompts | 1420 | 1416 |
| Epoch Capabilities Index | 142.41 | 141.29 |
| ForecastBench | 61.4 | 60 |
| SimpleBench | — | 40.8% |
| Kagi LLM Benchmark | — | 69.4% |
| NYT Connections (extended) | 14.3% | — |
| Chess Puzzles | 8% | — |
| LiveBench Reasoning | — | 83.2% |
| DTBench | 73.6% | — |
| LiveBench Data Analysis | — | 69.8% |
| LMCA | 30.9% | — |
| LiveBench | — | 71.6% |
Math Claude Haiku 4.5 leads
Claude Haiku 4.5: 44.9 (#78), DeepSeek-R1: 43.8 (#79)
| Benchmark | Claude Haiku 4.5 | DeepSeek-R1 |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 66.7% | 66.4% |
| Omni-MATH | 56.1% | 42.4% |
| LMArena Math | 1396 | 1400 |
| MATH Level 5 | 96.4% | 96.6% |
| LiveBench Math | — | 80.7% |
| FrontierMath (Feb 2025 set) | 5.9% | — |
| FrontierMath Tier 4 (v1) | 2.1% | — |
Knowledge DeepSeek-R1 leads
Claude Haiku 4.5: 37.7 (#153), DeepSeek-R1: 44.5 (#87)
| Benchmark | Claude Haiku 4.5 | DeepSeek-R1 |
|---|---|---|
| GPQA Diamond | 71.2% | 76.3% |
| MMLU-Pro | 77.7% | 79.3% |
| Vectara Hallucination Rate | 9.8% | 11.3% |
| GPQA (HELM) | 60.5% | 66.6% |
| LMArena Expert | 1442 | 1394 |
| SimpleQA Verified | 13.2% | — |
| Confabulations | — | 12.7% |
Multimodal Not comparable
Claude Haiku 4.5: 26.8 (#118), DeepSeek-R1: —
| Benchmark | Claude Haiku 4.5 | DeepSeek-R1 |
|---|---|---|
| Blueprint-Bench 2 | 0% | — |
| LMArena Document | 1420 | — |
Multilingual DeepSeek-R1 leads
Claude Haiku 4.5: 49.9 (#129), DeepSeek-R1: 52.4 (#85)
| Benchmark | Claude Haiku 4.5 | DeepSeek-R1 |
|---|---|---|
| LMArena Non-English | 1377 | 1412 |
| LMArena Chinese | 1417 | 1442 |
| LMArena French | 1408 | 1417 |
| LMArena German | 1375 | 1404 |
| LMArena Japanese | 1339 | 1391 |
| LMArena Korean | 1347 | 1360 |
| LMArena Russian | 1381 | 1423 |
| LMArena Spanish | 1420 | 1411 |
Instruction Following Too close to call
Claude Haiku 4.5: 71.4 (#149), DeepSeek-R1: 72.0 (#143)
| Benchmark | Claude Haiku 4.5 | DeepSeek-R1 |
|---|---|---|
| IFEval | 80.1% | 78.4% |
| LMArena Instruction Following | 1414 | 1382 |
| LiveBench Instruction Following | — | 80.5% |
Long Context DeepSeek-R1 leads
Claude Haiku 4.5: 43.6 (#92), DeepSeek-R1: 45.4 (#36)
| Benchmark | Claude Haiku 4.5 | DeepSeek-R1 |
|---|---|---|
| LMArena Longer Query | 1427 | 1391 |
| Fiction.LiveBench | — | 75% |
Writing & Preference DeepSeek-R1 leads
Claude Haiku 4.5: 57.9 (#123), DeepSeek-R1: 61.4 (#88)
| Benchmark | Claude Haiku 4.5 | DeepSeek-R1 |
|---|---|---|
| LMArena Text | 1396 | 1428 |
| LMArena Creative Writing | 1372 | 1405 |
| WildBench | 83.9% | 82.8% |
| LMArena Multi-Turn | 1409 | 1405 |
| Short-Story Creative Writing | — | 83% |
| EQ-Bench Creative Writing | — | 1500 |
| EQ-Bench 4 | 1064 | — |
| LiveBench Language | — | 48.5% |
Frequently asked questions
Is Claude Haiku 4.5 better than DeepSeek-R1?
DeepSeek-R1 is the stronger model overall, scoring 42.3 to 39.5 on the Noometry Index.
Which is cheaper, Claude Haiku 4.5 or DeepSeek-R1?
DeepSeek-R1 is cheaper. It lists at $0.50 per million input tokens and $2.15 per million output tokens; Claude Haiku 4.5 lists at $1 and $5.
Is Claude Haiku 4.5 or DeepSeek-R1 better for coding?
DeepSeek-R1 scores higher on coding benchmarks: 46.3 versus 44.0 in the Noometry coding category.
Which has the bigger context window?
Claude Haiku 4.5 does, with 200K tokens against 164K.
How many benchmarks do Claude Haiku 4.5 and DeepSeek-R1 share?
36 benchmarks have published results for both models. Claude Haiku 4.5 has 53 scored results on Noometry and DeepSeek-R1 has 52.