DeepSeek V4 Pro vs Inkling 2 Small
Full profiles: DeepSeek V4 Pro (DeepSeek) and Inkling 2 Small (Thinking Machines): every benchmark, price and rank.
Updated · Data from 8 sources
As of 6 Nov 2026, DeepSeek V4 Pro and Inkling 2 Small are neck and neck on quality (57.5 vs 57.4 on the BenchLeader Index, inside the ±6.6 margin), and Inkling 2 Small is 2.1× cheaper at $0.475 vs $0.990 per 1M tokens blended. Choose DeepSeek V4 Pro for reasoning (63 vs 53); choose Inkling 2 Small for agents & tools (69 vs 50) and coding (56 vs 54).
Key facts (as of 6 Nov 2026)
- DeepSeek V4 Pro scores 57.5 on the BenchLeader Index and Inkling 2 Small scores 57.4 (BenchLeader, 6 Nov 2026).
- DeepSeek V4 Pro costs $0.990 per 1M tokens blended (list price, 6 Nov 2026).
- Inkling 2 Small costs $0.475 per 1M tokens blended (list price, 6 Nov 2026).
DeepSeek V4 Pro benchmarks high DeepSeek Inkling 2 Small benchmarks Thinking MachinesAdd a third modelWhich to pick for…
The pick's numbers come first. Category scores on the BenchLeader scale (50 is the average model, 15 points one standard deviation). High confidence means a lead of half a standard deviation or more; under one point is called either.
| Use case | Pick | Why | Confidence |
|---|---|---|---|
| Coding | Inkling 2 Small | 56 vs 54 category score | Medium |
| Agents & tool use | Inkling 2 Small | 69 vs 50 category score | High |
| Reasoning | DeepSeek V4 Pro | 63 vs 53 category score | High |
| Chat quality (human preference) | – | Only DeepSeek V4 Pro has an Arena rating (65 category score) | Not comparable |
| Lowest cost | Inkling 2 Small | $0.475 vs $0.990 per 1M blended; agentic coding session $0.022 vs $0.048 | – |
| High volume on a budget | Neither flagship | A smaller model is usually the better fit: Cheapest good LLMs, Fastest LLMs | – |
Key differences
- DeepSeek V4 Pro (high) and Inkling 2 Small are neck and neck on quality (57.5 vs 57.4, within the ±6.6 margin).
- DeepSeek V4 Pro (high) is stronger in human preference, maths, reasoning.
- Inkling 2 Small is stronger in agents & tools, coding, knowledge, long context, multimodal.
- Inkling 2 Small is 2.1× cheaper ($0.475 vs $0.990 per 1M blended).
At a glance
| Spec | DeepSeek V4 Pro | Inkling 2 Small |
|---|---|---|
| Developer | DeepSeek | Thinking Machines |
| Released | 13 Aug 2026 | 5 Nov 2026 |
| Context window (tokens) | 1,048,576 | 1,024,000 |
| Compared at | DeepSeek V4 Pro (high) | Inkling 2 Small |
| Open weights | Yes | No |
| Price in / out / cached input per 1M | $0.660 / $1.98 / $0.022 | $0.300 / $1.00 / – |
Category by category
Category scores on the BenchLeader scale: 50 is the average evaluated model, 15 points is one standard deviation. A gap under one point is called level.
| Category | DeepSeek V4 Pro (high) | Inkling 2 Small | Leader |
|---|---|---|---|
| Reasoning | 62.5 | 52.5 | DeepSeek V4 Pro |
| Coding | 53.8 | 56.3 | Inkling 2 Small |
| Agents & tools | 50.2 | 69.3 | Inkling 2 Small |
| Maths | 64.5 | – | not comparable |
| Knowledge | – | 54.9 | not comparable |
| Human preference | 65.2 | – | not comparable |
| Multimodal | – | 64.1 | not comparable |
| Long context | – | 66.5 | not comparable |
On one chart
Each domain on the index's own scale, where 50 is an average model. Click a domain to see the benchmarks behind it. Pick up to 5 models from the list.
Not drawn, measured for fewer than two of these models: Maths, Knowledge, Instruction following, Long context, Multimodal, Human preference.
Where sources disagree
Across the 0 benchmarks both have, DeepSeek V4 Pro leads 0, Inkling 2 Small leads 0. Every overall index that rates both puts Inkling 2 Small ahead.
| Overall index | DeepSeek V4 Pro (high) | Inkling 2 Small | Leader |
|---|---|---|---|
| BenchLeader Index | 57.5 | 57.4 | neck and neck |
| AA Intelligence Index v5.0 | 12.6† | 30.6 | Inkling 2 Small |
| Vals Index | 53.0† | 56.1 | Inkling 2 Small |
Each publisher runs its own tasks, harnesses and effort settings, so their overall verdicts can differ; BenchLeader reconciles them on one scale. † measured at another effort setting of the same model (the publisher did not test the one compared here).
Price, speed and latency
List prices blended at three input tokens per output token. Per-task prices use the token counts on the methodology page. Cost, tokens and time per benchmark task are measured: Artificial Analysis’ averages over its whole Intelligence Index run.
| Measure | DeepSeek V4 Pro (high) | Inkling 2 Small |
|---|---|---|
| Blended price per 1M tokens | $0.990 | $0.475 |
| Input / output per 1M | $0.660 / $1.98 | $0.300 / $1.00 |
| Cached input per 1M | $0.022 | – |
| Price per task: chat reply | $0.0009 | $0.0004 |
| Price per task: summarize a 30-page report | $0.0091 | $0.0042 |
| Price per task: code edit | $0.0069 | $0.0033 |
| Price per task: agentic coding session | $0.048 | $0.022 |
| Price per task: structured extraction | $0.0017 | $0.0008 |
| Agentic coding session, input mostly cached | $0.019 | – |
| Cost per benchmark task (AA Index) | – | $1.18 |
| Output tokens per task (AA Index) | – | 150k |
| Of which reasoning | – | 88k |
| Time per task (AA Index) | – | 21 min |
| Output speed | 43 tok/s | – |
| Time to first answer | 1.3 s | – |
Each model is compared at its best-ranked reasoning-effort setting. DeepSeek V4 Pro (high) scores 57.5 and takes 1 s to its first answer token; DeepSeek V4 Pro (max) scores 57.0 in 26 s, DeepSeek V4 Pro scores 53.2 in 1 s. Inkling 2 Small scores 57.4. Higher effort means more thinking before the answer: better scores, longer waits and more tokens billed. How reasoning effort changes scores and cost.
Every benchmark
122 benchmarks, grouped by category; each name links to the board and its leaders. † measured at another effort setting of the same model, shown for reference and not counted as a win.
| Metric | Publisher | DeepSeek V4 Pro (high) | Inkling 2 Small |
|---|---|---|---|
| BenchLeader Index | BenchLeader | 57.5 | 57.4 |
| Agents & tools score | BenchLeader | 50.2 | 69.3 |
| Coding score | BenchLeader | 53.8 | 56.3 |
| Human preference score | BenchLeader | 65.2 | – |
| Maths score | BenchLeader | 64.5 | – |
| Reasoning score | BenchLeader | 62.5 | 52.5 |
| Knowledge score | BenchLeader | – | 54.9 |
| Long context score | BenchLeader | – | 66.5 |
| Multimodal score | BenchLeader | – | 64.1 |
| Blended price $/M | BenchLeader | $0.990 | $0.475 |
| Output speed | BenchLeader | 43 tok/s | – |
| Time to first answer | BenchLeader | 1.3 s | – |
| Context window | BenchLeader | 1,048,576 tokens | 1,024,000 tokens |
| Reasoning | |||
| GPQA Diamond | Epoch AI Benchmarking Hub | 90.9% | – |
| Arena Hard Prompts | Arena (LMArena) | 1483 | – |
| LiveBench 1.0 Reasoning | LiveBench | 85.8%† | – |
| GPQA Diamond (AA) | Artificial Analysis | 92.8%† | – |
| Humanity's Last Exam (AA) | Artificial Analysis | 41.0%† | 44.1% |
| HLE-Diamond (AA) | Artificial Analysis | 19.5%† | 28.7% |
| GPQA Diamond (Vals) | Vals AI | 92.4%† | – |
| Kagi LLM Benchmark | Kagi LLM Benchmark | 53.5%† | – |
| ARC-AGI-1 | ARC Prize | 87.2% | – |
| ARC-AGI-2 | ARC Prize | 59.7% | – |
| CritPt | Artificial Analysis | 18.0%† | – |
| CritPt v2 hard subset (AA) | Artificial Analysis | 23.5%† | 35.0% |
| Chess Puzzles | Epoch AI Benchmarking Hub | 16.0% | – |
| Mystery Game Puzzles | Epoch AI Benchmarking Hub | 43.0%† | – |
| CL-bench Life | Epoch AI Benchmarking Hub | 13.5% | – |
| LMCA | Epoch AI Benchmarking Hub | 45.5%† | – |
| DTBench | Epoch AI Benchmarking Hub | 93.9%† | – |
| ForecastBench | ForecastBench | 60.7%† | – |
| Coding | |||
| SWE-bench Verified (Epoch) | Epoch AI Benchmarking Hub | 77.6%† | – |
| SciCode | SciCode | 46.4% | – |
| WeirdML | WeirdML | 46.5% | – |
| FrontierCode | Cognition | 28.6% | – |
| Arena Coding | Arena (LMArena) | 1505 | – |
| Arena WebDev | Arena (LMArena) | 1582 | – |
| LiveBench 1.0 Coding | LiveBench | 77.2%† | – |
Show all 122 benchmarks
| Metric | Publisher | DeepSeek V4 Pro (high) | Inkling 2 Small |
|---|---|---|---|
| Coding | |||
| SciCode (AA) | Artificial Analysis | 51.0%† | 52.3% |
| LiveCodeBench | Vals AI | 87.5%† | – |
| IOI | Vals AI | 51.6%† | – |
| SWE-bench (Vals) | Vals AI | 96.4%† | – |
| Code Migration | Vals AI | 41.5%† | – |
| Code Migration v1.1 | Vals AI | 52.0%† | – |
| ProgramBench | Vals AI | 0.0%† | 0.5% |
| SWE-bench Multimodal | Vals AI | – | 43.1% |
| Vibe Code Bench 1-100 | Vals AI | 17.5%† | – |
| Vibe Code Bench v1.1 | Vals AI | 82.3%† | 81.5% |
| SWE Atlas: Codebase QnA | Scale AI SEAL | 27.1%† | – |
| SWE Atlas: Test Writing | Scale AI SEAL | 27.1%† | – |
| Surface Evolver Bench | Epoch AI Benchmarking Hub | 40.0% | – |
| ALE-Bench | Epoch AI Benchmarking Hub | 1006.1 | – |
| Terminal-Bench 4.0 (AA) | Artificial Analysis | 14.1%† | 13.1% |
| Terminal-Bench 2.1 (AA) | Artificial Analysis | 78.7%† | 81.3% |
| Startrise Frontend | Startrise | 62.7%† | – |
| GameBench 2 Overall | DuelLab GameBench 2 | 42.0† | – |
| GameBench 2 by reasoning setting | DuelLab GameBench 2 | 45.0 | – |
| FrontierCode 1.1 (AA) | Artificial Analysis | 38.8%† | – |
| FrontierCode 2.0 (AA) | Artificial Analysis | 17.1%† | 25.4% |
| Agents & tools | |||
| APEX-Agents | Mercor | 47.3%† | – |
| Arena Agent | Arena (LMArena) | -0.3 | – |
| LiveBench Agentic Coding | LiveBench | 55.0%† | – |
| Terminal-Bench 2.1 (Vals) | Vals AI | 50.2%† | – |
| HiL-Bench | Scale AI SEAL | 38.5%† | – |
| GDPval-AA v2.1, normalised score | Artificial Analysis | 47.5%† | – |
| GDPval-AA v2.1 | Artificial Analysis | 1457† | 1630 |
| τ³-Banking (AA) | Artificial Analysis | 39.6%† | 40.4% |
| Analyst Agent (AA) | Artificial Analysis | – | 51.3% |
| CyberBench | Vals AI | 72.9%† | 72.7% |
| Finance Agent v2 | Vals AI | 50.4%† | – |
| Finance Agent v2.1 | Vals AI | 57.1%† | 67.5% |
| Harvey's Legal Agent Benchmark v1.1 | Vals AI | 10.8%† | 21.7% |
| Harvey's Legal Agent Benchmark v1 | Vals AI | 7.5%† | – |
| Legal Research Bench | Vals AI | 40.9%† | 46.6% |
| SkillsBench | Vals AI | 53.8%† | – |
| SREBench | Vals AI | 1.9%† | 1.9% |
| Tax Agent Bench | Vals AI | 26.8%† | 38.9% |
| Terminal-Bench 2.0 (Vals) | Vals AI | 56.2%† | – |
| Terminal-Bench 4.0 (Vals) | Vals AI | 14.1%† | 12.6% |
| Terminal-Bench Science | Vals AI | 4.3%† | 2.9% |
| Vending-Bench 2 | Andon Labs | 3284.5† | – |
| AutomationBench | Artificial Analysis | 56.7%† | 45.6% |
| GDP.pdf (AA) | Artificial Analysis | 11.4%† | 25.2% |
| EnterpriseOps-Gym | Artificial Analysis | 49.6%† | 46.8% |
| AA-Briefcase v1.1 | Artificial Analysis | 1253† | 1527 |
| Terminal-Bench-Science 0.1 (AA) | Artificial Analysis | 5.7%† | 5.7% |
| Maths | |||
| FrontierMath Tiers 1–3 | Epoch AI Benchmarking Hub | 64.6%† | – |
| FrontierMath Tier 4 | Epoch AI Benchmarking Hub | 26.8%† | – |
| OTIS Mock AIME | Epoch AI Benchmarking Hub | 95.6% | – |
| ProofBench | Vals AI | 50.0%† | – |
| LiveBench Mathematics | LiveBench | 95.1%† | – |
| AIME 2026 | MathArena | 96.7%† | – |
| HMMT February 2026 | MathArena | 93.9%† | – |
| USAMO 2026 | MathArena | 60.7%† | – |
| MathArena Apex | MathArena | 28.1%† | – |
| Arena Maths | Arena (LMArena) | 1465 | – |
| Knowledge | |||
| SimpleQA Verified | Epoch AI Benchmarking Hub | 52.9%† | – |
| LiveBench 1.0 Data Analysis | LiveBench | 79.2%† | – |
| AA-Omniscience v1.1 | Artificial Analysis | -6.4† | 16.1 |
| MMLU-Pro | Vals AI | 87.3%† | – |
| LegalBench | Vals AI | 82.4%† | – |
| CorpFin | Vals AI | 65.4%† | – |
| TaxEval | Vals AI | 73.1%† | – |
| BioMysteryBench | Vals AI | – | 70.0% |
| CaseLaw v2 | Vals AI | 59.4%† | – |
| Excel Modeling Benchmark | Vals AI | 52.8%† | – |
| Excel Modeling Benchmark v1.1 | Vals AI | 65.1%† | 69.5% |
| MedCode | Vals AI | 42.5%† | – |
| MedScribe | Vals AI | 80.2%† | – |
| Public Benefits Bench v1.1 | Vals AI | 62.9%† | – |
| DrugDiscoveryBench | Scale AI SEAL | 31.7%† | – |
| AA-Omniscience v1.1: accuracy | Artificial Analysis | 45.2%† | 33.8% |
| AA-Omniscience v1.1: non-hallucination | Artificial Analysis | 5.9%† | 73.3% |
| Instruction following | |||
| LiveBench Language | LiveBench | 82.1%† | – |
| LiveBench Instruction Following | LiveBench | 67.7%† | – |
| Arena Instruction Following | Arena (LMArena) | 1462 | – |
| Human preference | |||
| Arena Text | Arena (LMArena) | 1465 | – |
| EQ-Bench 4 | EQ-Bench | 1156† | – |
| Arena Creative Writing | Arena (LMArena) | 1447 | – |
| Arena Multi-turn | Arena (LMArena) | 1471 | – |
| Arena Longer Queries | Arena (LMArena) | 1475 | – |
| Multimodal | |||
| MMMU-Pro | Artificial Analysis | – | 81.6% |
| SAGE | Vals AI | – | 88.6% |
| Code2Video-AA | Artificial Analysis | – | 1020 |
| AA-3D | Artificial Analysis | – | 1021 |
| Long context | |||
| AA-LCR v1.1 | Artificial Analysis | 80.3%† | 83.3% |
| Composite | |||
| Epoch Capabilities Index | Epoch AI Benchmarking Hub | 155.9† | – |
| LiveBench 1.0 | LiveBench | 77.4%† | – |
| AA Intelligence Index v5.0 | Artificial Analysis | 23.1† | 30.6 |
| AA Intelligence Index v4.3.2 | Artificial Analysis | 36† | – |
| AA Progress Index v1 | Artificial Analysis | 311.5† | – |
| Vals Index | Vals AI | 53.0† | 56.1 |
| Vals Finance Index | Vals AI | 52.9%† | 61.5% |
| Vals Router Index | Vals AI | 53.0%† | – |
| Metadata | |||
| AA Openness Index | Artificial Analysis | 44.4† | – |
Data as of 2026-11-06. Best configuration of each model; every score links to its source on the model pages.
DuelLab GameBench 2, release gb2-20261103-09b2218cbd85 (https://benchmarks.duellab.org/releases/gb2-20261103-09b2218cbd85/), CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). Scores mapped to BenchLeader model names; not endorsed by DuelLab.
What changed
- Inkling 2 Small: first results.
- DeepSeek V4 Pro (high): 1 score revision.
- DeepSeek V4 Pro (high): 1 score revision.
- DeepSeek V4 Pro (high): 16 new results (GPQA Diamond, OTIS Mock AIME, SciCode…); input price moved from $0.44/M to $0.66/M (50% up); output price moved from $0.87/M to $1.98/M (128% up).
- DeepSeek V4 Pro (high): BenchLeader Index 52.7 → 57.5.
- DeepSeek V4 Pro (high): 14 new results (AA Intelligence Index v4.3.2, GPQA Diamond (AA), Humanity's Last Exam (AA)…); input price moved from $0.66/M to $0.44/M (33% down); output price moved from $1.98/M to $0.87/M (56% down).
- DeepSeek V4 Pro (high): BenchLeader Index 55.8 → 52.7.
- DeepSeek V4 Pro (high): 1 new result (Arena WebDev); input price moved from $0.44/M to $0.66/M (50% up); output price moved from $0.87/M to $1.98/M (128% up).
- DeepSeek V4 Pro (high): 1 score revision.
- DeepSeek V4 Pro (high): 1 score revision.
Related comparisons
- Inkling 2 Small vs MiMo-V2.6-Pro
- Inkling 2 Small vs DeepSeek V4 Flash
- Inkling 2 Small vs Inkling Small
Rankings: Best LLM overall · Best LLM for agents · Best cheap LLMs
DeepSeek V4 Pro vs Inkling 2 Small: questions
- Is DeepSeek V4 Pro better than Inkling 2 Small?
- DeepSeek V4 Pro (high) and Inkling 2 Small are neck and neck on quality (57.5 vs 57.4, within the ±6.6 margin). The BenchLeader Index combines every independent quality benchmark; the two are within the margin of error as of 6 Nov 2026, so check the category scores for your use.
- DeepSeek V4 Pro vs Inkling 2 Small: which is cheaper, and will that change?
- Inkling 2 Small is cheaper today: $0.475 against $0.990 per million tokens, blended at three input tokens per output token. No price change has been announced for either.
- Is DeepSeek V4 Pro or Inkling 2 Small better for coding?
- Inkling 2 Small scores higher in coding (56 vs 54 on the category index, where 50 is average).
- Is DeepSeek V4 Pro or Inkling 2 Small better for agents and tool use?
- Inkling 2 Small scores higher in agentic tasks (69 vs 50 on the category index, where 50 is average).
- Which is better for long documents?
- Both take about 1M tokens: DeepSeek V4 Pro 1,048,576, Inkling 2 Small 1,024,000.
More questions (1)
- When was Inkling 2 Small released?
- Inkling 2 Small was released on 5 Nov 2026.