Model comparison
Claude Haiku 4.5 vs MiniMax-M3
MiniMax-M3 is the stronger model overall, scoring 43.5 to 39.3 on the Noometry Index.
Last verified . 32 shared benchmarks.
MiniMaxSummary
- They share 32 benchmarks with published results for both. Claude Haiku 4.5 scores higher in 3 categories and MiniMax-M3 in 7 categories; 9 gaps are clear of the uncertainty.
- The widest gap is in knowledge, where MiniMax-M3 leads 58.4 to 37.6.
- The biggest single-benchmark swing is NYT Connections (extended): 14.3% for Claude Haiku 4.5 and 65.1% for MiniMax-M3.
- MiniMax-M3 is cheaper at $0.30 / $1.20 per million input/output tokens, against $1 / $5 for Claude Haiku 4.5.
- MiniMax-M3 accepts more context: 1M tokens versus 200K.
- MiniMax-M3 has downloadable open weights; the other is API-only.
Side by side
| Claude Haiku 4.5 | MiniMax-M3 | |
|---|---|---|
| Provider | Anthropic | MiniMax |
| Noometry Index | 39.3 | 43.5 |
| Released | 2025-10-15 | 2026-06-01 |
| Weights | Proprietary | Open |
| Context window | 200K | 1M |
| Max output | 64K | 512K |
| Input $ / M tokens | $1 | $0.30 |
| Output $ / M tokens | $5 | $1.20 |
| Results tracked | 53 | 41 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Claude Haiku 4.5 leads
Claude Haiku 4.5: 43.9 (#71), MiniMax-M3: 41.5 (#111)
| Benchmark | Claude Haiku 4.5 | MiniMax-M3 |
|---|---|---|
| LMArena WebDev | 1330 | 1482 |
| SciCode | 43.3% | 47.1% |
| LMArena Coding | 1454 | 1470 |
| ALE-Bench | 653.48 | 640.02 |
| FrontierCode | — | 14.7% |
| SWE-bench Verified (bash only) | 66.6% | — |
| SWE-bench Multilingual | 64.7% | — |
| WeirdML | 45.4% | — |
Agentic & Tool Use Claude Haiku 4.5 leads
Claude Haiku 4.5: 33.5 (#50), MiniMax-M3: 22.5 (#121)
| Benchmark | Claude Haiku 4.5 | MiniMax-M3 |
|---|---|---|
| Vending-Bench 2 | 458.89 | 2,158 |
| Terminal-Bench | 35.5% | — |
| APEX-Agents | — | 37.7% |
| Berkeley Function Calling Leaderboard | 68.7% | — |
| OSWorld 2.0 | — | 4.6% |
| DeepResearch Bench | 45.5% | — |
| BALROG | 31.2% | — |
| ExploitBench | 13.7% | — |
| GBAEval | — | 0.9% |
Reasoning MiniMax-M3 leads
Claude Haiku 4.5: 15.1 (#310), MiniMax-M3: 29.3 (#83)
| Benchmark | Claude Haiku 4.5 | MiniMax-M3 |
|---|---|---|
| NYT Connections (extended) | 14.3% | 65.1% |
| CritPt | 0% | 3.7% |
| Chess Puzzles | 8% | 14% |
| LMArena Hard Prompts | 1420 | 1447 |
| DTBench | 73.6% | 78.9% |
| LMCA | 30.9% | 33.7% |
| Epoch Capabilities Index | 142.42 | 147 |
| ForecastBench | 61.4 | 61.4 |
| ARC-AGI-2 | 4% | — |
| SimpleBench | — | 45.8% |
| ARC-AGI-1 | 47.7% | — |
| Mystery Game Puzzles | — | 8% |
| Surface Evolver Bench | — | 55% |
Math Claude Haiku 4.5 leads
Claude Haiku 4.5: 43.0 (#69), MiniMax-M3: 39.1 (#83)
| Benchmark | Claude Haiku 4.5 | MiniMax-M3 |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 66.7% | 71.1% |
| LMArena Math | 1395 | 1432 |
| ProofBench | — | 18% |
| Omni-MATH | 56.1% | — |
| MATH Level 5 | 96.4% | — |
| FrontierMath (Feb 2025 set) | 5.9% | — |
| FrontierMath Tier 4 (v1) | 2.1% | — |
Knowledge MiniMax-M3 leads
Claude Haiku 4.5: 37.6 (#146), MiniMax-M3: 58.4 (#32)
| Benchmark | Claude Haiku 4.5 | MiniMax-M3 |
|---|---|---|
| GPQA Diamond | 71.2% | 90.9% |
| LMArena Expert | 1443 | 1462 |
| SimpleQA Verified | 13.2% | — |
| MMLU-Pro | 77.7% | — |
| Vectara Hallucination Rate | 9.8% | — |
| GPQA (HELM) | 60.5% | — |
Multimodal MiniMax-M3 leads
Claude Haiku 4.5: 27.0 (#111), MiniMax-M3: 39.5 (#45)
| Benchmark | Claude Haiku 4.5 | MiniMax-M3 |
|---|---|---|
| LMArena Document | 1420 | 1435 |
| LMArena Vision | — | 1254 |
| Blueprint-Bench 2 | 0% | — |
Multilingual MiniMax-M3 leads
Claude Haiku 4.5: 50.1 (#124), MiniMax-M3: 53.2 (#70)
| Benchmark | Claude Haiku 4.5 | MiniMax-M3 |
|---|---|---|
| LMArena Non-English | 1376 | 1420 |
| LMArena Chinese | 1415 | 1465 |
| LMArena French | 1408 | 1445 |
| LMArena German | 1377 | 1424 |
| LMArena Japanese | 1339 | 1381 |
| LMArena Korean | 1346 | 1372 |
| LMArena Russian | 1380 | 1430 |
| LMArena Spanish | 1420 | 1431 |
Instruction Following MiniMax-M3 leads
Claude Haiku 4.5: 71.6 (#144), MiniMax-M3: 75.7 (#56)
| Benchmark | Claude Haiku 4.5 | MiniMax-M3 |
|---|---|---|
| LMArena Instruction Following | 1414 | 1433 |
| IFEval | 80.1% | — |
Long Context Too close to call
Claude Haiku 4.5: 43.6 (#86), MiniMax-M3: 44.2 (#68)
| Benchmark | Claude Haiku 4.5 | MiniMax-M3 |
|---|---|---|
| LMArena Longer Query | 1427 | 1445 |
Writing & Preference MiniMax-M3 leads
Claude Haiku 4.5: 58.1 (#119), MiniMax-M3: 62.3 (#79)
| Benchmark | Claude Haiku 4.5 | MiniMax-M3 |
|---|---|---|
| LMArena Text | 1396 | 1433 |
| LMArena Creative Writing | 1371 | 1405 |
| EQ-Bench 4 | 1064 | 1150 |
| LMArena Multi-Turn | 1409 | 1441 |
| WildBench | 83.9% | — |
Frequently asked questions
Is Claude Haiku 4.5 better than MiniMax-M3?
MiniMax-M3 is the stronger model overall, scoring 43.5 to 39.3 on the Noometry Index.
Which is cheaper, Claude Haiku 4.5 or MiniMax-M3?
MiniMax-M3 is cheaper. It lists at $0.30 per million input tokens and $1.20 per million output tokens; Claude Haiku 4.5 lists at $1 and $5.
Is Claude Haiku 4.5 or MiniMax-M3 better for coding?
Claude Haiku 4.5 scores higher on coding benchmarks: 43.9 versus 41.5 in the Noometry coding category.
Which has the bigger context window?
MiniMax-M3 does, with 1M tokens against 200K.
How many benchmarks do Claude Haiku 4.5 and MiniMax-M3 share?
32 benchmarks have published results for both models. Claude Haiku 4.5 has 53 scored results on Noometry and MiniMax-M3 has 41.