Model comparison
Claude Haiku 4.5 vs Inkling
Inkling is the stronger model overall, scoring 44.1 to 39.5 on the Noometry Index.
Last verified . 32 shared benchmarks.
Summary
- They share 32 benchmarks with published results for both. Claude Haiku 4.5 scores higher in 3 categories and Inkling in 6 categories; 8 gaps are clear of the uncertainty.
- The widest gap is in reasoning, where Inkling leads 40.4 to 15.1.
- The biggest single-benchmark swing is ARC-AGI-2: 4% for Claude Haiku 4.5 and 36.5% for Inkling.
- Claude Haiku 4.5 is cheaper at $1 / $5 per million input/output tokens, against $1.87 / $4.68 for Inkling.
- Claude Haiku 4.5 accepts more context: 200K tokens versus 66K.
- Inkling has downloadable open weights; the other is API-only.
Side by side
| Claude Haiku 4.5 | Inkling | |
|---|---|---|
| Provider | Anthropic | Thinking Machines Lab |
| Noometry Index | 39.5 | 44.1 |
| Released | 2025-10-15 | 2026-07-15 |
| Weights | Proprietary | Open |
| Context window | 200K | 66K |
| Max output | 64K | 66K |
| Input $ / M tokens | $1 | $1.87 |
| Output $ / M tokens | $5 | $4.68 |
| 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: 44.0 (#78), Inkling: 34.5 (#234)
| Benchmark | Claude Haiku 4.5 | Inkling |
|---|---|---|
| LMArena WebDev | 1330 | 1413 |
| SciCode | 43.3% | 47% |
| WeirdML | 45.4% | 32.3% |
| LMArena Coding | 1453 | 1464 |
| ALE-Bench | 653.48 | 946 |
| FrontierCode | — | 14% |
| SWE-bench Verified (bash only) | 66.6% | — |
| SWE-bench Multilingual | 64.7% | — |
| FrontierSWE | — | 4.1% |
Agentic & Tool Use Claude Haiku 4.5 leads
Claude Haiku 4.5: 33.6 (#52), Inkling: 29.6 (#85)
| Benchmark | Claude Haiku 4.5 | Inkling |
|---|---|---|
| Terminal-Bench | 35.5% | — |
| APEX-Agents | — | 33.8% |
| Berkeley Function Calling Leaderboard | 68.7% | — |
| τ²-bench Banking | — | 25% |
| DeepResearch Bench | 45.5% | — |
| BALROG | 31.2% | — |
| ExploitBench | 13.7% | — |
| Vending-Bench 2 | 458.89 | — |
Reasoning Inkling leads
Claude Haiku 4.5: 15.1 (#320), Inkling: 40.4 (#56)
| Benchmark | Claude Haiku 4.5 | Inkling |
|---|---|---|
| ARC-AGI-2 | 4% | 36.5% |
| ARC-AGI-1 | 47.7% | 79.5% |
| CritPt | 0% | 5.4% |
| Chess Puzzles | 8% | 21% |
| LMArena Hard Prompts | 1420 | 1451 |
| DTBench | 73.6% | 87.5% |
| LMCA | 30.9% | 37.6% |
| Epoch Capabilities Index | 142.41 | 148.54 |
| SimpleBench | — | 50% |
| NYT Connections (extended) | 14.3% | — |
| ForecastBench | 61.4 | — |
Math Claude Haiku 4.5 leads
Claude Haiku 4.5: 44.9 (#78), Inkling: 31.3 (#225)
| Benchmark | Claude Haiku 4.5 | Inkling |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 66.7% | 88.9% |
| LMArena Math | 1396 | 1479 |
| FrontierMath (Tiers 1-3) | — | 33.3% |
| FrontierMath Tier 4 | — | 4.9% |
| ProofBench | — | 0% |
| Omni-MATH | 56.1% | — |
| MATH Level 5 | 96.4% | — |
| FrontierMath (Feb 2025 set) | 5.9% | — |
| FrontierMath Tier 4 (v1) | 2.1% | — |
Knowledge Inkling leads
Claude Haiku 4.5: 37.7 (#153), Inkling: 55.1 (#49)
| Benchmark | Claude Haiku 4.5 | Inkling |
|---|---|---|
| GPQA Diamond | 71.2% | 88.3% |
| SimpleQA Verified | 13.2% | 40.3% |
| LMArena Expert | 1442 | 1465 |
| MMLU-Pro | 77.7% | — |
| Vectara Hallucination Rate | 9.8% | — |
| GPQA (HELM) | 60.5% | — |
Multimodal Not comparable
Claude Haiku 4.5: 26.8 (#118), Inkling: —
| Benchmark | Claude Haiku 4.5 | Inkling |
|---|---|---|
| Blueprint-Bench 2 | 0% | — |
| LMArena Document | 1420 | — |
Multilingual Inkling leads
Claude Haiku 4.5: 49.9 (#129), Inkling: 54.0 (#52)
| Benchmark | Claude Haiku 4.5 | Inkling |
|---|---|---|
| LMArena Non-English | 1377 | 1434 |
| LMArena Chinese | 1417 | 1490 |
| LMArena French | 1408 | 1458 |
| LMArena German | 1375 | 1446 |
| LMArena Japanese | 1339 | 1429 |
| LMArena Korean | 1347 | 1404 |
| LMArena Russian | 1381 | 1429 |
| LMArena Spanish | 1420 | 1448 |
Instruction Following Inkling leads
Claude Haiku 4.5: 71.4 (#149), Inkling: 75.1 (#71)
| Benchmark | Claude Haiku 4.5 | Inkling |
|---|---|---|
| LMArena Instruction Following | 1414 | 1426 |
| IFEval | 80.1% | — |
Long Context Too close to call
Claude Haiku 4.5: 43.6 (#92), Inkling: 43.8 (#86)
| Benchmark | Claude Haiku 4.5 | Inkling |
|---|---|---|
| LMArena Longer Query | 1427 | 1434 |
Writing & Preference Inkling leads
Claude Haiku 4.5: 57.9 (#123), Inkling: 65.2 (#51)
| Benchmark | Claude Haiku 4.5 | Inkling |
|---|---|---|
| LMArena Text | 1396 | 1441 |
| LMArena Creative Writing | 1372 | 1387 |
| EQ-Bench 4 | 1064 | 1226 |
| LMArena Multi-Turn | 1409 | 1436 |
| EQ-Bench Creative Writing | — | 1611 |
| WildBench | 83.9% | — |
Frequently asked questions
Is Claude Haiku 4.5 better than Inkling?
Inkling is the stronger model overall, scoring 44.1 to 39.5 on the Noometry Index.
Which is cheaper, Claude Haiku 4.5 or Inkling?
Claude Haiku 4.5 is cheaper. It lists at $1 per million input tokens and $5 per million output tokens; Inkling lists at $1.87 and $4.68.
Is Claude Haiku 4.5 or Inkling better for coding?
Claude Haiku 4.5 scores higher on coding benchmarks: 44.0 versus 34.5 in the Noometry coding category.
Which has the bigger context window?
Claude Haiku 4.5 does, with 200K tokens against 66K.
How many benchmarks do Claude Haiku 4.5 and Inkling share?
32 benchmarks have published results for both models. Claude Haiku 4.5 has 53 scored results on Noometry and Inkling has 41.