Claude Haiku 4.5 vs Inkling
Updated October 21, 2026. 81 benchmarks have a result for both. Coding is ranked on the Vibe Coding Index; every other board is shown as published.
Decision reading
Claude Haiku 4.5 and Inkling are level for coding, 49.8 against 49.7 on the Vibe Coding Index, so price and context decide.
Pick Inkling when output tokens cost 1.2x less, or when the context window is larger, 524K against 200K tokens. Pick Claude Haiku 4.5 when a solved task costs $0.56 against $0.76, or when what matters is WeirdML or LiveBench browser, where it leads.
- Shared results
- 81
- Claude Haiku 4.5 only
- 13
- Inkling only
- 21
- Like-for-like categories
- 8 of 9
Shared results can support a head-to-head reading. Results only one model has describe coverage, not superiority.
Which one for your work
Each card is decided by one number both models carry. Where only one has it, the card says so and makes no call.
Coding
Vibe Coding Index, four public coding benchmarks
No call
Level: 49.8 against 49.7, inside 2 points.
3 and 2 benchmarks behind the scores
Intelligence
AA Intelligence Index
Inkling
Inkling leads, 25.0 against 16.9.
Same board, both measured
Agentic work
AA Agentic Index
Inkling
Inkling leads, 22.5 against 8.0.
Same board, both measured
Reasoning
GPQA Diamond
Inkling
Inkling leads, 87.1% against 72.2%.
Same board, both measured
Math
AIME
No call
Only Claude Haiku 4.5 has a AIME result, 82.7%.
Directional only
Knowledge
MMLU Pro
Inkling
Inkling leads, 86.3% against 78.7%.
Same board, both measured
Web design
Design Arena: websites
Inkling
Inkling leads, 1224 against 1122.
Same board, both measured
General
Vals Index
No call
Only Inkling has a Vals Index result, 29.0%.
Directional only
Long documents
Listed context window
Inkling
Inkling takes 524K tokens against 200K.
Listed by OpenRouter
Everyday cost
1K input + 500 output tokens
Inkling
Claude Haiku 4.5 $0.0035, Inkling $0.0030 per turn.
List prices
Cost per solved task
One SWE-bench Verified task, solved
Claude Haiku 4.5
Claude Haiku 4.5 $0.56, Inkling $0.76.
Published attempt cost and pass rate
At a glance
- Coding index
- Claude Haiku 4.5 49.8 from 3 benchmarks, Inkling 49.7 from 2 benchmarks.
- Widest category
- Reasoning, 16.9 points apart on average over 10 shared benchmarks.
- Widest single gap
- LiveCodeBench (Vals AI): Claude Haiku 4.5 41.2% against Inkling 85.5%, 44.3 points apart on a 0 to 100 scale.
- Output price
- Claude Haiku 4.5 $5.00 against Inkling $4.05 per million output tokens, 1.2x apart.
- Context window
- Claude Haiku 4.5 takes 200K tokens, Inkling 524K.
Benchmark by benchmark
First the score: boards each model leads, per category, where both have a result. Then every board, a category to a fold; bars are each result on a 0 to 100 scale, so compare the two in a row, not one row with another.
| Category | Claude Haiku 4.5 | Bars | Inkling | Ties |
|---|---|---|---|---|
| Coding | 1 | 12 | 1 tie | |
| Intelligence | 0 | 2 | ||
| Agentic work | 1 | 7 | ||
| Reasoning | 0 | 10 | ||
| Math | 0 | 3 | ||
| Knowledge | 0 | 11 | ||
| Web design | 2 | 7 | ||
| General | 0 | 24 | ||
| All shared boards | 4 | 76 | 1 tie |
CodingLike-for-likeFour of these boards feed the Vibe Coding Index112
| Benchmark | Claude Haiku 4.5 | Inkling | Leader |
|---|---|---|---|
| SWE-bench Verified | 66.6% | Not measured | Only Claude Haiku 4.5 measured |
| SWE-bench Multilingual | 64.7% | Not measured | Only Claude Haiku 4.5 measured |
| LiveBench coding | 48.1% | 64.0% | Inkling |
| LMArena WebDev | 1330 | 1414 | Inkling |
| AA Coding Index | 43.9 | 52.1 | Inkling |
| LiveCodeBench (Vals AI) | 41.2% | 85.5% | Inkling |
| SWE-bench Verified (Vals AI) | 66.6% | 77.6% | Inkling |
| Terminal-Bench 4 (Vals AI) | Not measured | 0.5% | Only Inkling measured |
| Vibe Code Bench (Vals AI) | 11.4% | 19.2% | Inkling |
| Code migration (Vals AI) | 7.5% | 11.8% | Inkling |
| ProgramBench (Vals AI) | 0.0% | 0.0% | Tie |
| IOI (Vals AI) | Not measured | 14.9% | Only Inkling measured |
| Terminal-Bench 2.1 (Vals AI) | 43.8% | 47.6% | Inkling |
| Vibe Code Bench 1-100 (Vals AI) | Not measured | 7.3% | Only Inkling measured |
| SkillsBench (Vals AI) | Not measured | 26.1% | Only Inkling measured |
| CyberBench (Vals AI) | Not measured | 66.8% | Only Inkling measured |
| SciCode | Not measured | 47.0% | Only Inkling measured |
| Terminal-Bench | 35.5% | Not measured | Only Claude Haiku 4.5 measured |
| FrontierCode | Not measured | 14.0% | Only Inkling measured |
| FrontierSWE | Not measured | 4.1% | Only Inkling measured |
| WeirdML | 45.4% | 32.3% | Claude Haiku 4.5 |
| ALE-Bench | Not measured | 946.0 | Only Inkling measured |
| LMArena Text: coding | 1454 | 1466 | Inkling |
| LMArena Text: software and IT | 1439 | 1465 | Inkling |
| LMArena WebDev: React | 1326 | 1404 | Inkling |
| LMArena WebDev: HTML | 1318 | 1399 | Inkling |
| Neither model has a result on the other 12 boards here | |||
IntelligenceLike-for-likePublic boards, shown as published, kept out of the index02
| Benchmark | Claude Haiku 4.5 | Inkling | Leader |
|---|---|---|---|
| AA Intelligence Index | 16.9 | 25.0 | Inkling |
| Epoch Capabilities Index | 142.2 | 148.4 | Inkling |
Agentic workLike-for-likePublic boards, shown as published, kept out of the index17
| Benchmark | Claude Haiku 4.5 | Inkling | Leader |
|---|---|---|---|
| AA Agentic Index | 8.0 | 22.5 | Inkling |
| Design Arena: agents building web apps | Not measured | 1196 | Only Inkling measured |
| LiveBench research | 42.2% | 59.9% | Inkling |
| LiveBench browser | 40.0% | 16.0% | Claude Haiku 4.5 |
| LiveBench long agentic | 11.1% | 14.1% | Inkling |
| SREBench (Vals AI) | Not measured | 0.0% | Only Inkling measured |
| Finance Agent (Vals AI) | 31.0% | 46.6% | Inkling |
| Tax Agent Bench (Vals AI) | 5.9% | 15.3% | Inkling |
| Legal Research Bench (Vals AI) | 10.6% | 28.4% | Inkling |
| Harvey's Legal Agent Benchmark (Vals AI) | Not measured | 3.8% | Only Inkling measured |
| Excel Modeling Benchmark (Vals AI) | 22.7% | 38.4% | Inkling |
| APEX-Agents | Not measured | 33.8% | Only Inkling measured |
| BALROG | 31.2% | Not measured | Only Claude Haiku 4.5 measured |
| DeepResearch Bench | 45.5% | Not measured | Only Claude Haiku 4.5 measured |
| Vending-Bench 2 | $459 | Not measured | Only Claude Haiku 4.5 measured |
| Neither model has a result on the other 24 boards here | |||
ReasoningLike-for-likePublic boards, shown as published, kept out of the index010
| Benchmark | Claude Haiku 4.5 | Inkling | Leader |
|---|---|---|---|
| GPQA Diamond | 72.2% | 87.1% | Inkling |
| LiveBench reasoning | 61.7% | 78.3% | Inkling |
| MMMU Pro (Vals AI) | 46.1% | 78.5% | Inkling |
| GPQA Diamond (Artificial Analysis run) | Not measured | 87.2% | Only Inkling measured |
| Humanity's Last Exam (Artificial Analysis run) | Not measured | 31.9% | Only Inkling measured |
| GPQA Diamond (Epoch AI run) | 71.2% | 88.3% | Inkling |
| ARC-AGI | 47.7% | 79.5% | Inkling |
| ARC-AGI-2 | 4.0% | 36.5% | Inkling |
| SimpleBench | Not measured | 50.0% | Only Inkling measured |
| Chess puzzles | 8.0% | 21.0% | Inkling |
| LMArena Text: hard prompts | 1421 | 1451 | Inkling |
| LMArena Text: expert | 1443 | 1466 | Inkling |
| LMArena Text: hard prompts in English | 1435 | 1448 | Inkling |
| Neither model has a result on the other 8 boards here | |||
MathLike-for-likePublic boards, shown as published, kept out of the index03
| Benchmark | Claude Haiku 4.5 | Inkling | Leader |
|---|---|---|---|
| AIME | 82.7% | Not measured | Only Claude Haiku 4.5 measured |
| MGSM (Vals AI) | 92.2% | Not measured | Only Claude Haiku 4.5 measured |
| OTIS Mock AIME | 66.7% | 88.9% | Inkling |
| FrontierMath, tiers 1 to 3 | Not measured | 33.3% | Only Inkling measured |
| FrontierMath, tier 4 | Not measured | 4.9% | Only Inkling measured |
| ProofBench | Not measured | 0.0% | Only Inkling measured |
| MATH level 5 | 96.4% | Not measured | Only Claude Haiku 4.5 measured |
| LMArena Text: math | 1398 | 1479 | Inkling |
| LMArena Text: mathematical work | 1430 | 1460 | Inkling |
| Neither model has a result on the other 4 boards here | |||
KnowledgeLike-for-likePublic boards, shown as published, kept out of the index011
| Benchmark | Claude Haiku 4.5 | Inkling | Leader |
|---|---|---|---|
| MMLU Pro | 78.7% | 86.3% | Inkling |
| LegalBench (Vals AI) | 81.2% | 83.0% | Inkling |
| CaseLaw (Vals AI) | 56.5% | Not measured | Only Claude Haiku 4.5 measured |
| CorpFin (Vals AI) | 60.6% | 68.6% | Inkling |
| TaxEval (Vals AI) | 67.5% | 75.3% | Inkling |
| MortgageTax (Vals AI) | 62.2% | 64.0% | Inkling |
| MedQA (Vals AI) | 79.6% | Not measured | Only Claude Haiku 4.5 measured |
| MedCode (Vals AI) | 32.7% | 41.2% | Inkling |
| SimpleQA Verified | 13.2% | 40.3% | Inkling |
| LMArena Text: science | 1406 | 1455 | Inkling |
| LMArena Text: medicine and healthcare | 1394 | 1447 | Inkling |
| LMArena Text: legal and government | 1400 | 1448 | Inkling |
| LMArena Text: business and finance | 1400 | 1438 | Inkling |
| Neither model has a result on the other 4 boards here | |||
Web designLike-for-likePublic boards, shown as published, kept out of the index27
| Benchmark | Claude Haiku 4.5 | Inkling | Leader |
|---|---|---|---|
| Design Arena: websites | 1122 | 1224 | Inkling |
| Design Arena: UI components | 1100 | 1200 | Inkling |
| Design Arena: game dev | 1097 | 1166 | Inkling |
| Design Arena: data viz | 1124 | 1160 | Inkling |
| Design Arena: 3D | 1072 | 1161 | Inkling |
| Design Arena: all code categories | 1115 | 1206 | Inkling |
| Design Arena: SVG | 1025 | 1109 | Inkling |
| Design Arena: ASCII art | 1139 | 1089 | Claude Haiku 4.5 |
| LiveBench design | 59.9% | 58.3% | Claude Haiku 4.5 |
GeneralLike-for-likePublic boards, shown as published, kept out of the index024
| Benchmark | Claude Haiku 4.5 | Inkling | Leader |
|---|---|---|---|
| Vals Index | Not measured | 29.0% | Only Inkling measured |
| LiveBench data analysis | 59.3% | 72.8% | Inkling |
| Public Benefits Bench (Vals AI) | 54.3% | 58.5% | Inkling |
| SAGE (Vals AI) | 31.8% | 36.5% | Inkling |
| MedScribe (Vals AI) | 85.2% | 85.4% | Inkling |
| Vals Multimodal Index | 42.9% | 49.4% | Inkling |
| Long context reasoning (AA-LCR) | Not measured | 77.3% | Only Inkling measured |
| DTBench | 73.6% | 87.5% | Inkling |
| LMCA | 30.9% | 37.6% | Inkling |
| ExploitBench | 13.7% | Not measured | Only Claude Haiku 4.5 measured |
| LMArena Text | 1397 | 1442 | Inkling |
| LMArena Text: instruction following | 1414 | 1425 | Inkling |
| LMArena Text: creative writing | 1371 | 1390 | Inkling |
| LMArena Text: English | 1414 | 1442 | Inkling |
| LMArena Text: non-English | 1377 | 1434 | Inkling |
| LMArena Text: Chinese | 1418 | 1493 | Inkling |
| LMArena Text: German | 1374 | 1445 | Inkling |
| LMArena Text: Japanese | 1338 | 1429 | Inkling |
| LMArena Text: Korean | 1346 | 1403 | Inkling |
| LMArena Text: French | 1411 | 1455 | Inkling |
| LMArena Text: Spanish | 1421 | 1449 | Inkling |
| LMArena Text: Russian | 1381 | 1430 | Inkling |
| LMArena Text: Polish | 1343 | 1450 | Inkling |
| LMArena Text: multi-turn | 1409 | 1437 | Inkling |
| LMArena Text: longer queries | 1427 | 1435 | Inkling |
| LMArena Text: writing and language | 1386 | 1404 | Inkling |
| LMArena Text: entertainment and media | 1369 | 1394 | Inkling |
| LMArena Document | 1420 | Not measured | Only Claude Haiku 4.5 measured |
| Neither model has a result on the other 18 boards here | |||
Image generationNot comparablePublic boards, shown as published, kept out of the index0 boards, none shared
| Benchmark | Claude Haiku 4.5 | Inkling | Leader |
|---|---|---|---|
| Neither model has a result on 5 boards here | |||
What each workload costs
Three fixed token mixes at each model's listed price, so per-token rates become a bill.
Chat turn
1K input + 500 output tokens
Inkling costs less
Repository review
50K input + 3K output tokens
Inkling costs less
Agent loop
200K input + 10K output tokens
Inkling costs less
Listed OpenRouter rates, without cache or batch discounts. A workload that does not fit a model's context window would need splitting.
Specification differences
The leading value in a row is marked; a row with a missing value names no leader. Our two rows are marked in red.
| Specification | Anthropic Claude Haiku 4.5 | Thinking Machines Inkling |
|---|---|---|
| Maker | Anthropic | Thinking Machines |
| Weights | No public weights listed | Open weights |
| Released | October 15, 2025 | July 17, 2026 |
| Context window | 200K | 524K |
| Input $/Mtok | $1.00 | $1.00 |
| Output $/Mtok | $5.00 | $4.05 |
| Coding index | 49.8 | 49.7 |
| Independent measurements behind it | 3 | 2 |
| Cost per solved task | $0.56 | $0.76 |
| Source | OpenRouter listing | OpenRouter listing |
Questions
Is Claude Haiku 4.5 better than Inkling for coding?
Claude Haiku 4.5 and Inkling are level for coding, 49.8 against 49.7 on the Vibe Coding Index, so price and context decide. Pick Inkling when output tokens cost 1.2x less, or when the context window is larger, 524K against 200K tokens. Pick Claude Haiku 4.5 when a solved task costs $0.56 against $0.76, or when what matters is WeirdML or LiveBench browser, where it leads.
Which is cheaper, Claude Haiku 4.5 or Inkling?
Claude Haiku 4.5 costs $1.00 per million input tokens and $5.00 per million output tokens; Inkling costs $1.00 and $4.05. Per solved SWE-bench Verified task, Claude Haiku 4.5 costs $0.56 and Inkling $0.76.
Which has the bigger context window, Claude Haiku 4.5 or Inkling?
Inkling, with 524K tokens against 200K.
How many benchmarks is the Claude Haiku 4.5 vs Inkling comparison based on?
81 benchmarks have a result for both models as of October 21, 2026. Claude Haiku 4.5's coding index rests on 3 independent benchmarks and Inkling's on 2. Where only one model has a result, the page says so instead of naming a winner.