UI-TARS 7B vs Gemma 4 31B
UI-TARS 7B costs 2.0x less per output token than Gemma 4 31B ($0.20 vs $0.40 per million). Updated October 22, 2026; 1 benchmarks have a result for both.
UI-TARS 7B or Gemma 4 31B: which to pick
Only Gemma 4 31B has a coding score, 39.0 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 2.0x cheaper ($0.20 vs $0.40 per million).
- Shared results
- 1
- UI-TARS 7B only
- 0
- Gemma 4 31B only
- 70
- 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 Gemma 4 31B has a coding score, so no call is made.
One side not measured
Intelligence
AA Intelligence Index
No call
Only Gemma 4 31B has a AA Intelligence Index result, 14.7.
Directional only
Agentic work
AA Agentic Index
No call
Only Gemma 4 31B has a AA Agentic Index result, 4.2.
Directional only
Long documents
Listed context window
Gemma 4 31B
Gemma 4 31B takes 262K tokens against 128K.
Listed by OpenRouter
Everyday cost
1K input + 500 output tokens
UI-TARS 7B
UI-TARS 7B $0.0002, Gemma 4 31B $0.0003 per turn.
List prices
At a glance
- Coding index
- UI-TARS 7B not benchmarked, Gemma 4 31B 39.0 from 2 benchmarks.
- Widest category
- Reasoning, 42.8 points apart on average over 1 shared benchmark.
- Widest single gap
- Kagi LLM Benchmark: UI-TARS 7B 20.7% against Gemma 4 31B 63.5%, 42.8 points apart on a 0 to 100 scale.
- Output price
- UI-TARS 7B $0.20 against Gemma 4 31B $0.40 per million output tokens, 2.0x apart.
- Context window
- UI-TARS 7B takes 128K tokens, Gemma 4 31B 262K.
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 | Gemma 4 31B | Ties |
|---|---|---|---|---|
| Reasoning | 0 | 1 | ||
| All shared boards | 0 | 1 |
CodingDirectional onlySix of these boards feed the Vibe Coding Index12 boards, none shared
| Benchmark | UI-TARS 7B | Gemma 4 31B | Leader |
|---|---|---|---|
| LiveBench coding | Not measured | 49.0% | Only Gemma 4 31B measured |
| LMArena WebDev | Not measured | 1367 | Only Gemma 4 31B measured |
| AA Coding Index | Not measured | 43.4 | Only Gemma 4 31B measured |
| SciCode | Not measured | 45.5% | Only Gemma 4 31B measured |
| Terminal-Bench Hard (Artificial Analysis run) | Not measured | 36.4% | Only Gemma 4 31B measured |
| WeirdML | Not measured | 52.3% | Only Gemma 4 31B measured |
| ALE-Bench | Not measured | 925.5 | Only Gemma 4 31B measured |
| LMArena Text: coding | Not measured | 1459 | Only Gemma 4 31B measured |
| LMArena Text: software and IT | Not measured | 1463 | Only Gemma 4 31B measured |
| LMArena WebDev: React | Not measured | 1359 | Only Gemma 4 31B measured |
| LMArena WebDev: HTML | Not measured | 1380 | Only Gemma 4 31B measured |
| SciCode | Not measured | 8.9% | Only Gemma 4 31B measured |
| Neither model has a result on the other 27 boards here | |||
IntelligenceDirectional onlyPublic boards, shown as published, kept out of the index2 boards, none shared
| Benchmark | UI-TARS 7B | Gemma 4 31B | Leader |
|---|---|---|---|
| AA Intelligence Index | Not measured | 14.7 | Only Gemma 4 31B measured |
| Epoch Capabilities Index | Not measured | 143.0 | Only Gemma 4 31B measured |
Agentic workDirectional onlyPublic boards, shown as published, kept out of the index5 boards, none shared
| Benchmark | UI-TARS 7B | Gemma 4 31B | Leader |
|---|---|---|---|
| AA Agentic Index | Not measured | 4.2 | Only Gemma 4 31B measured |
| LiveBench research | Not measured | 44.4% | Only Gemma 4 31B measured |
| LiveBench browser | Not measured | 4.0% | Only Gemma 4 31B measured |
| LiveBench long agentic | Not measured | 13.3% | Only Gemma 4 31B measured |
| τ²-Bench Telecom | Not measured | 65.5% | Only Gemma 4 31B measured |
| Neither model has a result on the other 36 boards here | |||
ReasoningLike-for-likePublic boards, shown as published, kept out of the index01
| Benchmark | UI-TARS 7B | Gemma 4 31B | Leader |
|---|---|---|---|
| LiveBench reasoning | Not measured | 68.3% | Only Gemma 4 31B measured |
| GPQA Diamond (Artificial Analysis run) | Not measured | 85.7% | Only Gemma 4 31B measured |
| Humanity's Last Exam (Artificial Analysis run) | Not measured | 23.6% | Only Gemma 4 31B measured |
| GPQA Diamond (Epoch AI run) | Not measured | 75.8% | Only Gemma 4 31B measured |
| LMArena Text: hard prompts | Not measured | 1448 | Only Gemma 4 31B measured |
| LMArena Text: expert | Not measured | 1464 | Only Gemma 4 31B measured |
| LMArena Text: hard prompts in English | Not measured | 1456 | Only Gemma 4 31B measured |
| LMArena Vision: diagrams | Not measured | 1283 | Only Gemma 4 31B measured |
| LMArena Vision: homework | Not measured | 1310 | Only Gemma 4 31B measured |
| Kagi LLM Benchmark | 20.7% | 63.5% | Gemma 4 31B |
| Neither model has a result on the other 12 boards here | |||
MathDirectional onlyPublic boards, shown as published, kept out of the index3 boards, none shared
| Benchmark | UI-TARS 7B | Gemma 4 31B | Leader |
|---|---|---|---|
| OTIS Mock AIME | Not measured | 73.3% | Only Gemma 4 31B measured |
| LMArena Text: math | Not measured | 1463 | Only Gemma 4 31B measured |
| LMArena Text: mathematical work | Not measured | 1465 | Only Gemma 4 31B measured |
| Neither model has a result on the other 10 boards here | |||
KnowledgeDirectional onlyPublic boards, shown as published, kept out of the index8 boards, none shared
| Benchmark | UI-TARS 7B | Gemma 4 31B | Leader |
|---|---|---|---|
| CaseLaw (Vals AI) | Not measured | 52.6% | Only Gemma 4 31B measured |
| MortgageTax (Vals AI) | Not measured | 61.4% | Only Gemma 4 31B measured |
| SimpleQA Verified | Not measured | 10.4% | Only Gemma 4 31B measured |
| LMArena Text: science | Not measured | 1457 | Only Gemma 4 31B measured |
| LMArena Text: medicine and healthcare | Not measured | 1451 | Only Gemma 4 31B measured |
| LMArena Text: legal and government | Not measured | 1438 | Only Gemma 4 31B measured |
| LMArena Text: business and finance | Not measured | 1430 | Only Gemma 4 31B measured |
| FACTS Benchmark Suite | Not measured | 45.3% | Only Gemma 4 31B measured |
| Neither model has a result on the other 10 boards here | |||
Web designDirectional onlyPublic boards, shown as published, kept out of the index1 board, none shared
| Benchmark | UI-TARS 7B | Gemma 4 31B | Leader |
|---|---|---|---|
| LiveBench design | Not measured | 58.4% | Only Gemma 4 31B measured |
| Neither model has a result on the other 8 boards here | |||
GeneralDirectional onlyPublic boards, shown as published, kept out of the index30 boards, none shared
| Benchmark | UI-TARS 7B | Gemma 4 31B | Leader |
|---|---|---|---|
| LiveBench data analysis | Not measured | 53.3% | Only Gemma 4 31B measured |
| SAGE (Vals AI) | Not measured | 55.0% | Only Gemma 4 31B measured |
| IFBench | Not measured | 75.6% | Only Gemma 4 31B measured |
| Long context reasoning (AA-LCR) | Not measured | 69.7% | Only Gemma 4 31B measured |
| DTBench | Not measured | 82.7% | Only Gemma 4 31B measured |
| LMCA | Not measured | 39.3% | Only Gemma 4 31B measured |
| Surface Evolver Bench | Not measured | 30.6% | Only Gemma 4 31B measured |
| LMArena Text | Not measured | 1443 | Only Gemma 4 31B measured |
| LMArena Text: instruction following | Not measured | 1433 | Only Gemma 4 31B measured |
| LMArena Text: creative writing | Not measured | 1417 | Only Gemma 4 31B measured |
| LMArena Vision | Not measured | 1276 | Only Gemma 4 31B measured |
| LMArena Text: English | Not measured | 1451 | Only Gemma 4 31B measured |
| LMArena Text: non-English | Not measured | 1430 | Only Gemma 4 31B measured |
| LMArena Text: Chinese | Not measured | 1475 | Only Gemma 4 31B measured |
| LMArena Text: French | Not measured | 1436 | Only Gemma 4 31B measured |
| LMArena Text: Spanish | Not measured | 1442 | Only Gemma 4 31B measured |
| LMArena Text: Russian | Not measured | 1460 | Only Gemma 4 31B measured |
| LMArena Text: Polish | Not measured | 1433 | Only Gemma 4 31B measured |
| LMArena Text: multi-turn | Not measured | 1452 | Only Gemma 4 31B measured |
| LMArena Text: longer queries | Not measured | 1446 | Only Gemma 4 31B measured |
| LMArena Text: writing and language | Not measured | 1432 | Only Gemma 4 31B measured |
| LMArena Text: entertainment and media | Not measured | 1405 | Only Gemma 4 31B measured |
| LMArena Vision: English | Not measured | 1271 | Only Gemma 4 31B measured |
| LMArena Vision: Chinese | Not measured | 1326 | Only Gemma 4 31B measured |
| LMArena Vision: reading text | Not measured | 1282 | Only Gemma 4 31B measured |
| LMArena Vision: humor | Not measured | 1280 | Only Gemma 4 31B measured |
| LMArena Vision: captioning | Not measured | 1218 | Only Gemma 4 31B measured |
| LMArena Vision: recognizing things | Not measured | 1226 | Only Gemma 4 31B measured |
| LMArena Vision: creative writing | Not measured | 1285 | Only Gemma 4 31B measured |
| LMArena Document | Not measured | 1425 | Only Gemma 4 31B measured |
| Neither model has a result on the other 16 boards here | |||
Image generationNot comparablePublic boards, shown as published, kept out of the index0 boards, none shared
| Benchmark | UI-TARS 7B | Gemma 4 31B | 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 | Google Gemma 4 31B |
|---|---|---|
| Maker | ByteDance | |
| Weights | Open weights | Open weights |
| Released | July 22, 2025 | April 2, 2026 |
| Context window | 128K | 262K |
| Input $/Mtok | $0.10 | $0.14 |
| Output $/Mtok | $0.20 | $0.40 |
| Coding index | Not benchmarked | 39.0 |
| Independent measurements behind it | 0 | 2 |
| Cost per solved task | Not computable | Not computable |
| Source | OpenRouter listing | OpenRouter listing |
Questions
Is UI-TARS 7B better than Gemma 4 31B for coding?
Only Gemma 4 31B has a coding score, 39.0 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 2.0x cheaper ($0.20 vs $0.40 per million).
Which is cheaper, UI-TARS 7B or Gemma 4 31B?
UI-TARS 7B costs $0.10 per million input tokens and $0.20 per million output tokens; Gemma 4 31B costs $0.14 and $0.40. Cost per solved task is known for neither of them.
Which has the bigger context window, UI-TARS 7B or Gemma 4 31B?
Gemma 4 31B, with 262K tokens against 128K.
How many benchmarks is the UI-TARS 7B vs Gemma 4 31B 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 Gemma 4 31B's on 2. Where only one model has a result, the page says so instead of naming a winner.