Enter opens the highlighted result. Esc closes.

UI-TARS 7B vs GPT-4

UI-TARS 7B costs 300.0x less per output token than GPT-4 ($0.20 vs $60.00 per million). Updated October 22, 2026; no benchmark has tested both models yet.

Replace UI-TARS 7B Replace GPT-4
UI-TARS 7BByteDance
Coding
No data
Overall
No data
Output $/M
$0.20
GPT-4OpenAI
Coding
No data
Overall
No data
Output $/M
$60.00
vs $0.20 vs $60.00 per 1M output tokens. UI-TARS 7B is 300.0x cheaper.

UI-TARS 7B or GPT-4: which to pick

Neither model has a coding score yet, so the two cannot be ranked on coding.

Pick UI-TARS 7B for output tokens 300.0x cheaper ($0.20 vs $60.00 per million) and a larger context window (128K against 8K tokens).

Model AUI-TARS 7BByteDanceNot benchmarkedNo coding benchmarkNo cost per solved taskModel BGPT-4OpenAINot benchmarkedNo coding benchmarkNo cost per solved task

No benchmark has tested both yet. UI-TARS 7B has results on Kagi LLM Benchmark; GPT-4 on SWE-bench Verified, SWE-bench Lite and SWE-bench (full test set). 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.

At a glance

Coding index
UI-TARS 7B not benchmarked, GPT-4 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
UI-TARS 7B $0.20 against GPT-4 $60.00 per million output tokens, 300.0x apart.
Context window
UI-TARS 7B takes 128K tokens, GPT-4 8K.

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 Index6 boards measured
Coding benchmarks for UI-TARS 7B and GPT-4
BenchmarkUI-TARS 7BGPT-4Leader
SWE-bench VerifiedNot measured22.4%Only GPT-4 measured
SWE-bench LiteNot measured39.7%Only GPT-4 measured
SWE-bench (full test set)Not measured12.5%Only GPT-4 measured
AA Coding IndexNot measured13.1Only GPT-4 measured
LMArena Text: codingNot measured1209Only GPT-4 measured
LMArena Text: software and ITNot measured1195Only GPT-4 measured
Neither model has a result on the other 33 boards here
IntelligenceDirectional onlyPublic boards, shown as published, kept out of the index2 boards measured
Intelligence benchmarks for UI-TARS 7B and GPT-4
BenchmarkUI-TARS 7BGPT-4Leader
AA Intelligence IndexNot measured6.7Only GPT-4 measured
Epoch Capabilities IndexNot measured125.7Only GPT-4 measured
ReasoningDirectional onlyPublic boards, shown as published, kept out of the index6 boards measured
Reasoning benchmarks for UI-TARS 7B and GPT-4
BenchmarkUI-TARS 7BGPT-4Leader
GPQA Diamond (Artificial Analysis run)Not measured34.9%Only GPT-4 measured
BIG-Bench HardNot measured75.1%Only GPT-4 measured
LMArena Text: hard promptsNot measured1200Only GPT-4 measured
LMArena Text: expertNot measured1149Only GPT-4 measured
LMArena Text: hard prompts in EnglishNot measured1209Only GPT-4 measured
Kagi LLM Benchmark20.7%Not measuredOnly UI-TARS 7B measured
Neither model has a result on the other 16 boards here
MathDirectional onlyPublic boards, shown as published, kept out of the index3 boards measured
Math benchmarks for UI-TARS 7B and GPT-4
BenchmarkUI-TARS 7BGPT-4Leader
GSM8KNot measured92.0%Only GPT-4 measured
LMArena Text: mathNot measured1230Only GPT-4 measured
LMArena Text: mathematical workNot measured1228Only GPT-4 measured
Neither model has a result on the other 10 boards here
KnowledgeDirectional onlyPublic boards, shown as published, kept out of the index7 boards measured
Knowledge benchmarks for UI-TARS 7B and GPT-4
BenchmarkUI-TARS 7BGPT-4Leader
MMLU Pro (Artificial Analysis run)Not measured56.2%Only GPT-4 measured
MMLUNot measured86.4%Only GPT-4 measured
AI2 Reasoning ChallengeNot measured96.3%Only GPT-4 measured
LMArena Text: scienceNot measured1177Only GPT-4 measured
LMArena Text: medicine and healthcareNot measured1145Only GPT-4 measured
LMArena Text: legal and governmentNot measured1191Only GPT-4 measured
LMArena Text: business and financeNot measured1145Only GPT-4 measured
Neither model has a result on the other 11 boards here
GeneralDirectional onlyPublic boards, shown as published, kept out of the index22 boards measured
General benchmarks for UI-TARS 7B and GPT-4
BenchmarkUI-TARS 7BGPT-4Leader
IFBenchNot measured33.2%Only GPT-4 measured
DTBenchNot measured62.7%Only GPT-4 measured
LMCANot measured17.1%Only GPT-4 measured
HellaSwagNot measured95.3%Only GPT-4 measured
WinoGrandeNot measured87.5%Only GPT-4 measured
LMArena TextNot measured1206Only GPT-4 measured
LMArena Text: instruction followingNot measured1201Only GPT-4 measured
LMArena Text: creative writingNot measured1191Only GPT-4 measured
LMArena Text: EnglishNot measured1220Only GPT-4 measured
LMArena Text: non-EnglishNot measured1194Only GPT-4 measured
LMArena Text: ChineseNot measured1184Only GPT-4 measured
LMArena Text: GermanNot measured1198Only GPT-4 measured
LMArena Text: JapaneseNot measured1137Only GPT-4 measured
LMArena Text: KoreanNot measured1087Only GPT-4 measured
LMArena Text: FrenchNot measured1220Only GPT-4 measured
LMArena Text: SpanishNot measured1197Only GPT-4 measured
LMArena Text: RussianNot measured1195Only GPT-4 measured
LMArena Text: PolishNot measured1265Only GPT-4 measured
LMArena Text: multi-turnNot measured1206Only GPT-4 measured
LMArena Text: longer queriesNot measured1190Only GPT-4 measured
LMArena Text: writing and languageNot measured1219Only GPT-4 measured
LMArena Text: entertainment and mediaNot measured1184Only GPT-4 measured
Neither model has a result on the other 24 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$0.0002Fits in one request
GPT-4$0.060Fits in one request

UI-TARS 7B costs less

Repository review

50K input + 3K output tokens

UI-TARS 7B$0.0056Fits in one request
GPT-4$1.68Exceeds the context window

UI-TARS 7B costs less

Agent loop

200K input + 10K output tokens

UI-TARS 7B$0.022Exceeds the context window
GPT-4$6.60Exceeds the context window

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.

Specifications of UI-TARS 7B and GPT-4
SpecificationByteDance UI-TARS 7BOpenAI GPT-4
MakerByteDanceOpenAI
WeightsOpen weightsNo public weights listed
ReleasedJuly 22, 2025May 28, 2023
Context window128K8K
Input $/Mtok$0.10$30.00
Output $/Mtok$0.20$60.00
Coding indexNot benchmarkedNot benchmarked
Independent measurements behind it00
Cost per solved taskNot computableNot computable
SourceOpenRouter listingOpenRouter listing

Questions

Is UI-TARS 7B better than GPT-4 for coding?

Neither model has a coding score yet, so the two cannot be ranked on coding. Pick UI-TARS 7B for output tokens 300.0x cheaper ($0.20 vs $60.00 per million) and a larger context window (128K against 8K tokens).

Which is cheaper, UI-TARS 7B or GPT-4?

UI-TARS 7B costs $0.10 per million input tokens and $0.20 per million output tokens; GPT-4 costs $30.00 and $60.00. Cost per solved task is known for neither of them.

Which has the bigger context window, UI-TARS 7B or GPT-4?

UI-TARS 7B, with 128K tokens against 8K.

Has any benchmark tested both UI-TARS 7B and GPT-4?

No benchmark has tested both yet. UI-TARS 7B has results on Kagi LLM Benchmark; GPT-4 on SWE-bench Verified, SWE-bench Lite and SWE-bench (full test set). Compare them on price and context instead. Data as of October 22, 2026.

Compare next