Inkling vs Voxtral Small 24B 2507
Voxtral Small 24B 2507 costs 13.5x less per output token than Inkling ($0.30 vs $4.05 per million). Updated October 21, 2026; no benchmark has tested both models yet.
- Coding
- 31.5
- Overall
- 46.5
- Output $/M
- $4.05
- Coding
- No data
- Overall
- No data
- Output $/M
- $0.30
Inkling or Voxtral Small 24B 2507: which to pick
Only Inkling has a coding score, 31.5 on the Vibe Coding Index; Voxtral Small 24B 2507 has none, so the two cannot be ranked on coding.
Pick Voxtral Small 24B 2507 for output tokens 13.5x cheaper ($0.30 vs $4.05 per million).
No benchmark has tested both yet. Inkling has results on LiveBench coding, LMArena WebDev and AA Coding Index; Voxtral Small 24B 2507 has none yet. Compare them on price and context instead.
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, six public coding benchmarks
No call
Only Inkling has a coding score, so no call is made.
One side not measured
Intelligence
AA Intelligence Index
No call
Only Inkling has a AA Intelligence Index result, 25.0.
Directional only
Agentic work
AA Agentic Index
No call
Only Inkling has a AA Agentic Index result, 22.5.
Directional only
Reasoning
GPQA Diamond
No call
Only Inkling has a GPQA Diamond result, 87.1%.
Directional only
Knowledge
MMLU Pro
No call
Only Inkling has a MMLU Pro result, 86.3%.
Directional only
Web design
Design Arena: websites
No call
Only Inkling has a Design Arena: websites result, 1224.
Directional only
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 33K.
Listed by OpenRouter
Everyday cost
1K input + 500 output tokens
Voxtral Small 24B 2507
Inkling $0.0030, Voxtral Small 24B 2507 $0.0003 per turn.
List prices
Cost per solved task
One SWE-bench Verified task, solved
No call
Only Inkling has the number, $0.76.
Directional only
At a glance
- Coding index
- Inkling 31.5 from 6 benchmarks, Voxtral Small 24B 2507 not benchmarked.
- Widest category
- None yet: no category has been tested on both models.
- Widest single gap
- None yet: no benchmark has tested both models.
- Output price
- Inkling $4.05 against Voxtral Small 24B 2507 $0.30 per million output tokens, 13.5x apart.
- Context window
- Inkling takes 524K tokens, Voxtral Small 24B 2507 33K.
Benchmark by benchmark
Every board either model has a result on, a category to a fold; bars are each result on a 0 to 100 scale. Only one model is measured on each, so no row names a leader.
CodingDirectional onlySix of these boards feed the Vibe Coding Index23 boards measured
| Benchmark | Inkling | Voxtral Small 24B 2507 | Leader |
|---|---|---|---|
| LiveBench coding | 64.0% | Not measured | Only Inkling measured |
| LMArena WebDev | 1414 | Not measured | Only Inkling measured |
| AA Coding Index | 52.1 | Not measured | Only Inkling measured |
| LiveCodeBench (Vals AI) | 85.5% | Not measured | Only Inkling measured |
| SWE-bench Verified (Vals AI) | 77.6% | Not measured | Only Inkling measured |
| Terminal-Bench 4 (Vals AI) | 0.5% | Not measured | Only Inkling measured |
| Vibe Code Bench (Vals AI) | 19.2% | Not measured | Only Inkling measured |
| Code migration (Vals AI) | 11.8% | Not measured | Only Inkling measured |
| ProgramBench (Vals AI) | 0.0% | Not measured | Only Inkling measured |
| IOI (Vals AI) | 14.9% | Not measured | Only Inkling measured |
| Terminal-Bench 2.1 (Vals AI) | 47.6% | Not measured | Only Inkling measured |
| Vibe Code Bench 1-100 (Vals AI) | 7.3% | Not measured | Only Inkling measured |
| SkillsBench (Vals AI) | 26.1% | Not measured | Only Inkling measured |
| CyberBench (Vals AI) | 66.8% | Not measured | Only Inkling measured |
| SciCode | 47.0% | Not measured | Only Inkling measured |
| FrontierCode | 14.0% | Not measured | Only Inkling measured |
| FrontierSWE | 4.1% | Not measured | Only Inkling measured |
| WeirdML | 32.3% | Not measured | Only Inkling measured |
| ALE-Bench | 946.0 | Not measured | Only Inkling measured |
| LMArena Text: coding | 1466 | Not measured | Only Inkling measured |
| LMArena Text: software and IT | 1465 | Not measured | Only Inkling measured |
| LMArena WebDev: React | 1404 | Not measured | Only Inkling measured |
| LMArena WebDev: HTML | 1399 | Not measured | Only Inkling measured |
| Neither model has a result on the other 16 boards here | |||
IntelligenceDirectional onlyPublic boards, shown as published, kept out of the index2 boards measured
| Benchmark | Inkling | Voxtral Small 24B 2507 | Leader |
|---|---|---|---|
| AA Intelligence Index | 25.0 | Not measured | Only Inkling measured |
| Epoch Capabilities Index | 148.4 | Not measured | Only Inkling measured |
Agentic workDirectional onlyPublic boards, shown as published, kept out of the index12 boards measured
| Benchmark | Inkling | Voxtral Small 24B 2507 | Leader |
|---|---|---|---|
| AA Agentic Index | 22.5 | Not measured | Only Inkling measured |
| Design Arena: agents building web apps | 1196 | Not measured | Only Inkling measured |
| LiveBench research | 59.9% | Not measured | Only Inkling measured |
| LiveBench browser | 16.0% | Not measured | Only Inkling measured |
| LiveBench long agentic | 14.1% | Not measured | Only Inkling measured |
| SREBench (Vals AI) | 0.0% | Not measured | Only Inkling measured |
| Finance Agent (Vals AI) | 46.6% | Not measured | Only Inkling measured |
| Tax Agent Bench (Vals AI) | 15.3% | Not measured | Only Inkling measured |
| Legal Research Bench (Vals AI) | 28.4% | Not measured | Only Inkling measured |
| Harvey's Legal Agent Benchmark (Vals AI) | 3.8% | Not measured | Only Inkling measured |
| Excel Modeling Benchmark (Vals AI) | 38.4% | Not measured | Only Inkling measured |
| APEX-Agents | 33.8% | Not measured | Only Inkling measured |
| Neither model has a result on the other 29 boards here | |||
ReasoningDirectional onlyPublic boards, shown as published, kept out of the index13 boards measured
| Benchmark | Inkling | Voxtral Small 24B 2507 | Leader |
|---|---|---|---|
| GPQA Diamond | 87.1% | Not measured | Only Inkling measured |
| LiveBench reasoning | 78.3% | Not measured | Only Inkling measured |
| MMMU Pro (Vals AI) | 78.5% | Not measured | Only Inkling measured |
| GPQA Diamond (Artificial Analysis run) | 87.2% | Not measured | Only Inkling measured |
| Humanity's Last Exam (Artificial Analysis run) | 31.9% | Not measured | Only Inkling measured |
| GPQA Diamond (Epoch AI run) | 88.3% | Not measured | Only Inkling measured |
| ARC-AGI | 79.5% | Not measured | Only Inkling measured |
| ARC-AGI-2 | 36.5% | Not measured | Only Inkling measured |
| SimpleBench | 50.0% | Not measured | Only Inkling measured |
| Chess puzzles | 21.0% | Not measured | Only Inkling measured |
| LMArena Text: hard prompts | 1451 | Not measured | Only Inkling measured |
| LMArena Text: expert | 1466 | Not measured | Only Inkling measured |
| LMArena Text: hard prompts in English | 1448 | Not measured | Only Inkling measured |
| Neither model has a result on the other 9 boards here | |||
MathDirectional onlyPublic boards, shown as published, kept out of the index6 boards measured
| Benchmark | Inkling | Voxtral Small 24B 2507 | Leader |
|---|---|---|---|
| OTIS Mock AIME | 88.9% | Not measured | Only Inkling measured |
| FrontierMath, tiers 1 to 3 | 33.3% | Not measured | Only Inkling measured |
| FrontierMath, tier 4 | 4.9% | Not measured | Only Inkling measured |
| ProofBench | 0.0% | Not measured | Only Inkling measured |
| LMArena Text: math | 1479 | Not measured | Only Inkling measured |
| LMArena Text: mathematical work | 1460 | Not measured | Only Inkling measured |
| Neither model has a result on the other 7 boards here | |||
KnowledgeDirectional onlyPublic boards, shown as published, kept out of the index11 boards measured
| Benchmark | Inkling | Voxtral Small 24B 2507 | Leader |
|---|---|---|---|
| MMLU Pro | 86.3% | Not measured | Only Inkling measured |
| LegalBench (Vals AI) | 83.0% | Not measured | Only Inkling measured |
| CorpFin (Vals AI) | 68.6% | Not measured | Only Inkling measured |
| TaxEval (Vals AI) | 75.3% | Not measured | Only Inkling measured |
| MortgageTax (Vals AI) | 64.0% | Not measured | Only Inkling measured |
| MedCode (Vals AI) | 41.2% | Not measured | Only Inkling measured |
| SimpleQA Verified | 40.3% | Not measured | Only Inkling measured |
| LMArena Text: science | 1455 | Not measured | Only Inkling measured |
| LMArena Text: medicine and healthcare | 1447 | Not measured | Only Inkling measured |
| LMArena Text: legal and government | 1448 | Not measured | Only Inkling measured |
| LMArena Text: business and finance | 1438 | Not measured | Only Inkling measured |
| Neither model has a result on the other 7 boards here | |||
Web designDirectional onlyPublic boards, shown as published, kept out of the index9 boards measured
| Benchmark | Inkling | Voxtral Small 24B 2507 | Leader |
|---|---|---|---|
| Design Arena: websites | 1224 | Not measured | Only Inkling measured |
| Design Arena: UI components | 1201 | Not measured | Only Inkling measured |
| Design Arena: game dev | 1166 | Not measured | Only Inkling measured |
| Design Arena: data viz | 1159 | Not measured | Only Inkling measured |
| Design Arena: 3D | 1161 | Not measured | Only Inkling measured |
| Design Arena: all code categories | 1206 | Not measured | Only Inkling measured |
| Design Arena: SVG | 1110 | Not measured | Only Inkling measured |
| Design Arena: ASCII art | 1089 | Not measured | Only Inkling measured |
| LiveBench design | 58.3% | Not measured | Only Inkling measured |
GeneralDirectional onlyPublic boards, shown as published, kept out of the index26 boards measured
| Benchmark | Inkling | Voxtral Small 24B 2507 | Leader |
|---|---|---|---|
| Vals Index | 29.0% | Not measured | Only Inkling measured |
| LiveBench data analysis | 72.8% | Not measured | Only Inkling measured |
| Public Benefits Bench (Vals AI) | 58.5% | Not measured | Only Inkling measured |
| SAGE (Vals AI) | 36.5% | Not measured | Only Inkling measured |
| MedScribe (Vals AI) | 85.4% | Not measured | Only Inkling measured |
| Vals Multimodal Index | 49.4% | Not measured | Only Inkling measured |
| Long context reasoning (AA-LCR) | 77.3% | Not measured | Only Inkling measured |
| DTBench | 87.5% | Not measured | Only Inkling measured |
| LMCA | 37.6% | Not measured | Only Inkling measured |
| LMArena Text | 1442 | Not measured | Only Inkling measured |
| LMArena Text: instruction following | 1425 | Not measured | Only Inkling measured |
| LMArena Text: creative writing | 1390 | Not measured | Only Inkling measured |
| LMArena Text: English | 1442 | Not measured | Only Inkling measured |
| LMArena Text: non-English | 1434 | Not measured | Only Inkling measured |
| LMArena Text: Chinese | 1493 | Not measured | Only Inkling measured |
| LMArena Text: German | 1445 | Not measured | Only Inkling measured |
| LMArena Text: Japanese | 1429 | Not measured | Only Inkling measured |
| LMArena Text: Korean | 1403 | Not measured | Only Inkling measured |
| LMArena Text: French | 1455 | Not measured | Only Inkling measured |
| LMArena Text: Spanish | 1449 | Not measured | Only Inkling measured |
| LMArena Text: Russian | 1430 | Not measured | Only Inkling measured |
| LMArena Text: Polish | 1450 | Not measured | Only Inkling measured |
| LMArena Text: multi-turn | 1437 | Not measured | Only Inkling measured |
| LMArena Text: longer queries | 1435 | Not measured | Only Inkling measured |
| LMArena Text: writing and language | 1404 | Not measured | Only Inkling measured |
| LMArena Text: entertainment and media | 1394 | Not measured | Only Inkling measured |
| Neither model has a result on the other 20 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
Voxtral Small 24B 2507 costs less
Repository review
50K input + 3K output tokens
Voxtral Small 24B 2507 costs less
Agent loop
200K input + 10K output tokens
Voxtral Small 24B 2507 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 | Thinking Machines Inkling | Mistral Voxtral Small 24B 2507 |
|---|---|---|
| Maker | Thinking Machines | Mistral AI |
| Weights | Open weights | Open weights |
| Released | July 17, 2026 | October 30, 2025 |
| Context window | 524K | 33K |
| Input $/Mtok | $0.975 | $0.10 |
| Output $/Mtok | $4.05 | $0.30 |
| Coding index | 31.5 | Not benchmarked |
| Independent measurements behind it | 6 | 0 |
| Cost per solved task | $0.76 | Not computable |
| Source | OpenRouter listing | OpenRouter listing |
Questions
Is Inkling better than Voxtral Small 24B 2507 for coding?
Only Inkling has a coding score, 31.5 on the Vibe Coding Index; Voxtral Small 24B 2507 has none, so the two cannot be ranked on coding. Pick Voxtral Small 24B 2507 for output tokens 13.5x cheaper ($0.30 vs $4.05 per million).
Which is cheaper, Inkling or Voxtral Small 24B 2507?
Inkling costs $0.975 per million input tokens and $4.05 per million output tokens; Voxtral Small 24B 2507 costs $0.10 and $0.30. Cost per solved task is known for Inkling only.
Which has the bigger context window, Inkling or Voxtral Small 24B 2507?
Inkling, with 524K tokens against 33K.
Has any benchmark tested both Inkling and Voxtral Small 24B 2507?
No benchmark has tested both yet. Inkling has results on LiveBench coding, LMArena WebDev and AA Coding Index; Voxtral Small 24B 2507 has none yet. Compare them on price and context instead. Data as of October 21, 2026.