UI-TARS 7B vs Claude Sonnet 4.5
UI-TARS 7B costs 75.0x less per output token than Claude Sonnet 4.5 ($0.20 vs $15.00 per million). Updated October 22, 2026; 1 benchmarks have a result for both.
UI-TARS 7B or Claude Sonnet 4.5: which to pick
Only Claude Sonnet 4.5 has a coding score, 39.5 on the Vibe Coding Index; UI-TARS 7B has none, so the two cannot be ranked on coding.
Pick UI-TARS 7B for output tokens 75.0x cheaper ($0.20 vs $15.00 per million).
- Shared results
- 1
- UI-TARS 7B only
- 0
- Claude Sonnet 4.5 only
- 99
- Like-for-like categories
- 1 of 9
Only benchmarks that tested both models count toward a pick; a result only one model has is shown but not compared.
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 Claude Sonnet 4.5 has a coding score, so no call is made.
One side not measured
Intelligence
AA Intelligence Index
No call
Only Claude Sonnet 4.5 has a AA Intelligence Index result, 20.7.
Directional only
Agentic work
AA Agentic Index
No call
Only Claude Sonnet 4.5 has a AA Agentic Index result, 15.8.
Directional only
Reasoning
GPQA Diamond
No call
Only Claude Sonnet 4.5 has a GPQA Diamond result, 81.6%.
Directional only
Math
AIME
No call
Only Claude Sonnet 4.5 has a AIME result, 88.2%.
Directional only
Knowledge
MMLU Pro
No call
Only Claude Sonnet 4.5 has a MMLU Pro result, 87.4%.
Directional only
Web design
Design Arena: websites
No call
Only Claude Sonnet 4.5 has a Design Arena: websites result, 1189.
Directional only
Long documents
Listed context window
Claude Sonnet 4.5
Claude Sonnet 4.5 takes 1M tokens against 128K.
Listed by OpenRouter
Everyday cost
1K input + 500 output tokens
UI-TARS 7B
UI-TARS 7B $0.0002, Claude Sonnet 4.5 $0.011 per turn.
List prices
Cost per solved task
One SWE-bench Verified task, solved
No call
Only Claude Sonnet 4.5 has the number, $1.16.
Directional only
At a glance
- Coding index
- UI-TARS 7B not benchmarked, Claude Sonnet 4.5 39.5 from 3 benchmarks.
- Widest category
- Reasoning, 37.2 points apart on average over 1 shared benchmark.
- Widest single gap
- Kagi LLM Benchmark: UI-TARS 7B 20.7% against Claude Sonnet 4.5 57.9%, 37.2 points apart on a 0 to 100 scale.
- Output price
- UI-TARS 7B $0.20 against Claude Sonnet 4.5 $15.00 per million output tokens, 75.0x apart.
- Context window
- UI-TARS 7B takes 128K tokens, Claude Sonnet 4.5 1M.
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 | UI-TARS 7B | Bars | Claude Sonnet 4.5 | Ties |
|---|---|---|---|---|
| Reasoning | 0 | 1 | ||
| All shared boards | 0 | 1 |
CodingDirectional onlySix of these boards feed the Vibe Coding Index15 boards, none shared
| Benchmark | UI-TARS 7B | Claude Sonnet 4.5 | Leader |
|---|---|---|---|
| SWE-bench Verified | Not measured | 74.8% | Only Claude Sonnet 4.5 measured |
| SWE-bench Multilingual | Not measured | 67.0% | Only Claude Sonnet 4.5 measured |
| LMArena WebDev | Not measured | 1394 | Only Claude Sonnet 4.5 measured |
| AA Coding Index | Not measured | 52.1 | Only Claude Sonnet 4.5 measured |
| LiveCodeBench (Vals AI) | Not measured | 73.0% | Only Claude Sonnet 4.5 measured |
| SWE-bench Verified (Vals AI) | Not measured | 70.0% | Only Claude Sonnet 4.5 measured |
| Vibe Code Bench (Vals AI) | Not measured | 22.6% | Only Claude Sonnet 4.5 measured |
| Terminal-Bench | Not measured | 46.5% | Only Claude Sonnet 4.5 measured |
| GSO-Bench | Not measured | 14.7% | Only Claude Sonnet 4.5 measured |
| WeirdML | Not measured | 47.7% | Only Claude Sonnet 4.5 measured |
| Cybench | Not measured | 60.0% | Only Claude Sonnet 4.5 measured |
| LMArena Text: coding | Not measured | 1488 | Only Claude Sonnet 4.5 measured |
| LMArena Text: software and IT | Not measured | 1473 | Only Claude Sonnet 4.5 measured |
| LMArena WebDev: React | Not measured | 1386 | Only Claude Sonnet 4.5 measured |
| LMArena WebDev: HTML | Not measured | 1399 | Only Claude Sonnet 4.5 measured |
| Neither model has a result on the other 28 boards here | |||
IntelligenceDirectional onlyPublic boards, shown as published, kept out of the index2 boards, none shared
| Benchmark | UI-TARS 7B | Claude Sonnet 4.5 | Leader |
|---|---|---|---|
| AA Intelligence Index | Not measured | 20.7 | Only Claude Sonnet 4.5 measured |
| Epoch Capabilities Index | Not measured | 146.7 | Only Claude Sonnet 4.5 measured |
Agentic workDirectional onlyPublic boards, shown as published, kept out of the index12 boards, none shared
| Benchmark | UI-TARS 7B | Claude Sonnet 4.5 | Leader |
|---|---|---|---|
| AA Agentic Index | Not measured | 15.8 | Only Claude Sonnet 4.5 measured |
| Design Arena: agents building web apps | Not measured | 1060 | Only Claude Sonnet 4.5 measured |
| Design Arena: agents building full-stack apps | Not measured | 1050 | Only Claude Sonnet 4.5 measured |
| Design Arena: agents building mobile apps | Not measured | 1136 | Only Claude Sonnet 4.5 measured |
| METR time horizons | Not measured | 67.4% | Only Claude Sonnet 4.5 measured |
| Remote Labor Index | Not measured | 2.1% | Only Claude Sonnet 4.5 measured |
| GDPval | Not measured | 42.5% | Only Claude Sonnet 4.5 measured |
| DeepResearch Bench | Not measured | 52.6% | Only Claude Sonnet 4.5 measured |
| OSWorld | Not measured | 62.9% | Only Claude Sonnet 4.5 measured |
| Vending-Bench 2 | Not measured | $3,839 | Only Claude Sonnet 4.5 measured |
| Berkeley Function Calling Leaderboard (BFCL) | Not measured | 73.2% | Only Claude Sonnet 4.5 measured |
| MCP Atlas | Not measured | 59.5% | Only Claude Sonnet 4.5 measured |
| Neither model has a result on the other 30 boards here | |||
ReasoningLike-for-likePublic boards, shown as published, kept out of the index01
| Benchmark | UI-TARS 7B | Claude Sonnet 4.5 | Leader |
|---|---|---|---|
| GPQA Diamond | Not measured | 81.6% | Only Claude Sonnet 4.5 measured |
| MMMU Pro (Vals AI) | Not measured | 79.3% | Only Claude Sonnet 4.5 measured |
| GPQA Diamond (Epoch AI run) | Not measured | 82.3% | Only Claude Sonnet 4.5 measured |
| ARC-AGI | Not measured | 63.7% | Only Claude Sonnet 4.5 measured |
| ARC-AGI-2 | Not measured | 13.6% | Only Claude Sonnet 4.5 measured |
| SimpleBench | Not measured | 54.3% | Only Claude Sonnet 4.5 measured |
| Chess puzzles | Not measured | 10.0% | Only Claude Sonnet 4.5 measured |
| Mystery game puzzles | Not measured | 17.0% | Only Claude Sonnet 4.5 measured |
| EBR-bench | Not measured | 2.4% | Only Claude Sonnet 4.5 measured |
| VPCT | Not measured | 39.8% | Only Claude Sonnet 4.5 measured |
| LMArena Text: hard prompts | Not measured | 1462 | Only Claude Sonnet 4.5 measured |
| LMArena Text: expert | Not measured | 1483 | Only Claude Sonnet 4.5 measured |
| LMArena Text: hard prompts in English | Not measured | 1471 | Only Claude Sonnet 4.5 measured |
| Kagi LLM Benchmark | 20.7% | 57.9% | Claude Sonnet 4.5 |
| Neither model has a result on the other 9 boards here | |||
MathDirectional onlyPublic boards, shown as published, kept out of the index10 boards, none shared
| Benchmark | UI-TARS 7B | Claude Sonnet 4.5 | Leader |
|---|---|---|---|
| AIME | Not measured | 88.2% | Only Claude Sonnet 4.5 measured |
| MGSM (Vals AI) | Not measured | 94.3% | Only Claude Sonnet 4.5 measured |
| OTIS Mock AIME | Not measured | 77.8% | Only Claude Sonnet 4.5 measured |
| FrontierMath, tiers 1 to 3 | Not measured | 23.9% | Only Claude Sonnet 4.5 measured |
| FrontierMath, tier 4 | Not measured | 2.4% | Only Claude Sonnet 4.5 measured |
| ProofBench | Not measured | 19.0% | Only Claude Sonnet 4.5 measured |
| MATH level 5 | Not measured | 97.7% | Only Claude Sonnet 4.5 measured |
| LMArena Text: math | Not measured | 1449 | Only Claude Sonnet 4.5 measured |
| LMArena Text: mathematical work | Not measured | 1462 | Only Claude Sonnet 4.5 measured |
| MathArena | Not measured | 12.6% | Only Claude Sonnet 4.5 measured |
| Neither model has a result on the other 4 boards here | |||
KnowledgeDirectional onlyPublic boards, shown as published, kept out of the index16 boards, none shared
| Benchmark | UI-TARS 7B | Claude Sonnet 4.5 | Leader |
|---|---|---|---|
| MMLU Pro | Not measured | 87.4% | Only Claude Sonnet 4.5 measured |
| LegalBench (Vals AI) | Not measured | 84.1% | Only Claude Sonnet 4.5 measured |
| CaseLaw (Vals AI) | Not measured | 62.2% | Only Claude Sonnet 4.5 measured |
| CorpFin (Vals AI) | Not measured | 62.0% | Only Claude Sonnet 4.5 measured |
| TaxEval (Vals AI) | Not measured | 73.3% | Only Claude Sonnet 4.5 measured |
| MortgageTax (Vals AI) | Not measured | 64.0% | Only Claude Sonnet 4.5 measured |
| MedQA (Vals AI) | Not measured | 94.7% | Only Claude Sonnet 4.5 measured |
| MedCode (Vals AI) | Not measured | 44.1% | Only Claude Sonnet 4.5 measured |
| SimpleQA Verified | Not measured | 30.7% | Only Claude Sonnet 4.5 measured |
| LMArena Text: science | Not measured | 1448 | Only Claude Sonnet 4.5 measured |
| LMArena Text: medicine and healthcare | Not measured | 1444 | Only Claude Sonnet 4.5 measured |
| LMArena Text: legal and government | Not measured | 1451 | Only Claude Sonnet 4.5 measured |
| LMArena Text: business and finance | Not measured | 1440 | Only Claude Sonnet 4.5 measured |
| FACTS Benchmark Suite | Not measured | 40.6% | Only Claude Sonnet 4.5 measured |
| PRBench Finance | Not measured | 43.8% | Only Claude Sonnet 4.5 measured |
| PRBench Legal | Not measured | 40.8% | Only Claude Sonnet 4.5 measured |
| Neither model has a result on the other 4 boards here | |||
Web designDirectional onlyPublic boards, shown as published, kept out of the index8 boards, none shared
| Benchmark | UI-TARS 7B | Claude Sonnet 4.5 | Leader |
|---|---|---|---|
| Design Arena: websites | Not measured | 1189 | Only Claude Sonnet 4.5 measured |
| Design Arena: UI components | Not measured | 1170 | Only Claude Sonnet 4.5 measured |
| Design Arena: game dev | Not measured | 1166 | Only Claude Sonnet 4.5 measured |
| Design Arena: data viz | Not measured | 1167 | Only Claude Sonnet 4.5 measured |
| Design Arena: 3D | Not measured | 1157 | Only Claude Sonnet 4.5 measured |
| Design Arena: all code categories | Not measured | 1182 | Only Claude Sonnet 4.5 measured |
| Design Arena: SVG | Not measured | 1114 | Only Claude Sonnet 4.5 measured |
| Design Arena: ASCII art | Not measured | 1205 | Only Claude Sonnet 4.5 measured |
| Neither model has a result on the other 1 board here | |||
GeneralDirectional onlyPublic boards, shown as published, kept out of the index23 boards, none shared
| Benchmark | UI-TARS 7B | Claude Sonnet 4.5 | Leader |
|---|---|---|---|
| SAGE (Vals AI) | Not measured | 36.1% | Only Claude Sonnet 4.5 measured |
| MedScribe (Vals AI) | Not measured | 84.5% | Only Claude Sonnet 4.5 measured |
| DTBench | Not measured | 83.2% | Only Claude Sonnet 4.5 measured |
| LMCA | Not measured | 38.8% | Only Claude Sonnet 4.5 measured |
| LMArena Text | Not measured | 1438 | Only Claude Sonnet 4.5 measured |
| LMArena Text: instruction following | Not measured | 1459 | Only Claude Sonnet 4.5 measured |
| LMArena Text: creative writing | Not measured | 1442 | Only Claude Sonnet 4.5 measured |
| LMArena Text: English | Not measured | 1450 | Only Claude Sonnet 4.5 measured |
| LMArena Text: non-English | Not measured | 1425 | Only Claude Sonnet 4.5 measured |
| LMArena Text: Chinese | Not measured | 1460 | Only Claude Sonnet 4.5 measured |
| LMArena Text: German | Not measured | 1427 | Only Claude Sonnet 4.5 measured |
| LMArena Text: Japanese | Not measured | 1390 | Only Claude Sonnet 4.5 measured |
| LMArena Text: Korean | Not measured | 1402 | Only Claude Sonnet 4.5 measured |
| LMArena Text: French | Not measured | 1459 | Only Claude Sonnet 4.5 measured |
| LMArena Text: Spanish | Not measured | 1458 | Only Claude Sonnet 4.5 measured |
| LMArena Text: Russian | Not measured | 1437 | Only Claude Sonnet 4.5 measured |
| LMArena Text: Polish | Not measured | 1423 | Only Claude Sonnet 4.5 measured |
| LMArena Text: multi-turn | Not measured | 1465 | Only Claude Sonnet 4.5 measured |
| LMArena Text: longer queries | Not measured | 1475 | Only Claude Sonnet 4.5 measured |
| LMArena Text: writing and language | Not measured | 1442 | Only Claude Sonnet 4.5 measured |
| LMArena Text: entertainment and media | Not measured | 1428 | Only Claude Sonnet 4.5 measured |
| LMArena Document | Not measured | 1450 | Only Claude Sonnet 4.5 measured |
| MultiChallenge | Not measured | 55.3% | Only Claude Sonnet 4.5 measured |
| Neither model has a result on the other 25 boards here | |||
Image generationNot comparablePublic boards, shown as published, kept out of the index0 boards, none shared
| Benchmark | UI-TARS 7B | Claude Sonnet 4.5 | 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
UI-TARS 7B costs less
Repository review
50K input + 3K output tokens
UI-TARS 7B costs less
Agent loop
200K input + 10K output tokens
UI-TARS 7B 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 | ByteDance UI-TARS 7B | Anthropic Claude Sonnet 4.5 |
|---|---|---|
| Maker | ByteDance | Anthropic |
| Weights | Open weights | No public weights listed |
| Released | July 22, 2025 | September 29, 2025 |
| Context window | 128K | 1M |
| Input $/Mtok | $0.10 | $3.00 |
| Output $/Mtok | $0.20 | $15.00 |
| Coding index | Not benchmarked | 39.5 |
| Independent measurements behind it | 0 | 3 |
| Cost per solved task | Not computable | $1.16 |
| Source | OpenRouter listing | OpenRouter listing |
Questions
Is UI-TARS 7B better than Claude Sonnet 4.5 for coding?
Only Claude Sonnet 4.5 has a coding score, 39.5 on the Vibe Coding Index; UI-TARS 7B has none, so the two cannot be ranked on coding. Pick UI-TARS 7B for output tokens 75.0x cheaper ($0.20 vs $15.00 per million).
Which is cheaper, UI-TARS 7B or Claude Sonnet 4.5?
UI-TARS 7B costs $0.10 per million input tokens and $0.20 per million output tokens; Claude Sonnet 4.5 costs $3.00 and $15.00. Cost per solved task is known for Claude Sonnet 4.5 only.
Which has the bigger context window, UI-TARS 7B or Claude Sonnet 4.5?
Claude Sonnet 4.5, with 1M tokens against 128K.
How many benchmarks is the UI-TARS 7B vs Claude Sonnet 4.5 comparison based on?
1 benchmarks have a result for both models as of October 22, 2026. UI-TARS 7B's coding index rests on 0 independent benchmarks and Claude Sonnet 4.5's on 3. Where only one model has a result, the page says so instead of naming a winner.