Enter opens the highlighted result. Esc closes.

gpt-oss-20b vs Llama 3.3 70B Instruct

gpt-oss-20b leads Llama 3.3 70B Instruct on the Artificial Analysis Intelligence Index, 10.0 to 7.7, and gpt-oss-20b costs 4.8x less per output token ($0.15 vs $0.72 per million). Updated October 22, 2026; 44 benchmarks have a result for both.

Replace gpt-oss-20b Replace Llama 3.3 70B Instruct
gpt-oss-20bOpenAI
Coding
No data
Overall
38.5
Output $/M
$0.15
33
Llama 3.3 70B InstructMeta Llama
Coding
No data
Overall
26.0
Output $/M
$0.72
11
vs boards won of 44 shared

gpt-oss-20b or Llama 3.3 70B Instruct: which to pick

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

Pick gpt-oss-20b for output tokens 4.8x cheaper ($0.15 vs $0.72 per million) and work measured by LMArena Text: legal and government or LMCA, where it leads.

Model Agpt-oss-20bOpenAINot benchmarkedNo coding benchmarkNo cost per solved taskModel BLlama 3.3 70B InstructMeta LlamaNot benchmarkedNo coding benchmarkNo cost per solved task
Shared results
44
gpt-oss-20b only
18
Llama 3.3 70B Instruct only
8
Like-for-like categories
7 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.

At a glance

Coding index
gpt-oss-20b not benchmarked, Llama 3.3 70B Instruct not benchmarked.
Widest category
Math, 47.6 points apart on average over 5 shared benchmarks.
Widest single gap
AIME 2025 (Artificial Analysis run): gpt-oss-20b 89.3% against Llama 3.3 70B Instruct 7.7%, 81.7 points apart on a 0 to 100 scale.
Output price
gpt-oss-20b $0.15 against Llama 3.3 70B Instruct $0.72 per million output tokens, 4.8x apart.
Context window
Both take 131K tokens.

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.

Boards won, where both have a result
Categorygpt-oss-20bBarsLlama 3.3 70B InstructTies
Coding50
Intelligence20
Agentic work10
Reasoning60
Math50
Knowledge41
General1010
All shared boards3311
CodingLike-for-likeSix of these boards feed the Vibe Coding Index50
Coding benchmarks for gpt-oss-20b and Llama 3.3 70B Instruct
Benchmarkgpt-oss-20bLlama 3.3 70B InstructLeader
AA Coding IndexNot measured11.9Only Llama 3.3 70B Instruct measured
LiveCodeBench (Vals AI)80.4%Not measuredOnly gpt-oss-20b measured
LiveCodeBench (Artificial Analysis run)77.7%28.8%gpt-oss-20b
SciCode38.9%Not measuredOnly gpt-oss-20b measured
Terminal-Bench Hard (Artificial Analysis run)10.6%3.0%gpt-oss-20b
Terminal-Bench3.4%Not measuredOnly gpt-oss-20b measured
WeirdML40.9%14.4%gpt-oss-20b
ALE-Bench566.0Not measuredOnly gpt-oss-20b measured
LMArena Text: coding13071268gpt-oss-20b
LMArena Text: software and IT13111273gpt-oss-20b
Neither model has a result on the other 29 boards here
IntelligenceLike-for-likePublic boards, shown as published, kept out of the index20
Intelligence benchmarks for gpt-oss-20b and Llama 3.3 70B Instruct
Benchmarkgpt-oss-20bLlama 3.3 70B InstructLeader
AA Intelligence Index10.07.7gpt-oss-20b
Epoch Capabilities Index137.7127.3gpt-oss-20b
Agentic workLike-for-likePublic boards, shown as published, kept out of the index10
Agentic work benchmarks for gpt-oss-20b and Llama 3.3 70B Instruct
Benchmarkgpt-oss-20bLlama 3.3 70B InstructLeader
τ²-Bench Telecom60.2%26.6%gpt-oss-20b
BALROGNot measured23.0%Only Llama 3.3 70B Instruct measured
Berkeley Function Calling Leaderboard (BFCL)Not measured31.9%Only Llama 3.3 70B Instruct measured
BrowseComp-Plus50.6%Not measuredOnly gpt-oss-20b measured
Neither model has a result on the other 37 boards here
ReasoningLike-for-likePublic boards, shown as published, kept out of the index60
Reasoning benchmarks for gpt-oss-20b and Llama 3.3 70B Instruct
Benchmarkgpt-oss-20bLlama 3.3 70B InstructLeader
GPQA Diamond68.9%Not measuredOnly gpt-oss-20b measured
GPQA Diamond (Artificial Analysis run)68.8%49.8%gpt-oss-20b
Humanity's Last Exam (Artificial Analysis run)11.0%3.6%gpt-oss-20b
GPQA Diamond (Epoch AI run)60.8%47.4%gpt-oss-20b
SimpleBenchNot measured19.9%Only Llama 3.3 70B Instruct measured
LMArena Text: hard prompts12741257gpt-oss-20b
LMArena Text: expert12581225gpt-oss-20b
LMArena Text: hard prompts in English12921277gpt-oss-20b
Kagi LLM Benchmark53.2%Not measuredOnly gpt-oss-20b measured
Neither model has a result on the other 13 boards here
MathLike-for-likePublic boards, shown as published, kept out of the index50
Math benchmarks for gpt-oss-20b and Llama 3.3 70B Instruct
Benchmarkgpt-oss-20bLlama 3.3 70B InstructLeader
AIME86.0%Not measuredOnly gpt-oss-20b measured
MATH 500 (Vals AI)94.2%Not measuredOnly gpt-oss-20b measured
MGSM (Vals AI)89.0%Not measuredOnly gpt-oss-20b measured
AA Math Index89.37.7gpt-oss-20b
AIME 2025 (Artificial Analysis run)89.3%7.7%gpt-oss-20b
OTIS Mock AIME65.3%5.1%gpt-oss-20b
MATH level 5Not measured41.6%Only Llama 3.3 70B Instruct measured
LMArena Text: math13171267gpt-oss-20b
LMArena Text: mathematical work13051267gpt-oss-20b
Neither model has a result on the other 4 boards here
KnowledgeLike-for-likePublic boards, shown as published, kept out of the index41
Knowledge benchmarks for gpt-oss-20b and Llama 3.3 70B Instruct
Benchmarkgpt-oss-20bLlama 3.3 70B InstructLeader
MMLU Pro71.6%Not measuredOnly gpt-oss-20b measured
LegalBench (Vals AI)70.8%Not measuredOnly gpt-oss-20b measured
CaseLaw (Vals AI)43.8%Not measuredOnly gpt-oss-20b measured
CorpFin (Vals AI)53.1%Not measuredOnly gpt-oss-20b measured
TaxEval (Vals AI)63.7%Not measuredOnly gpt-oss-20b measured
MedQA (Vals AI)82.9%Not measuredOnly gpt-oss-20b measured
MMLU Pro (Artificial Analysis run)74.8%71.3%gpt-oss-20b
MMLUNot measured86.3%Only Llama 3.3 70B Instruct measured
LMArena Text: science12961284gpt-oss-20b
LMArena Text: medicine and healthcare13101275gpt-oss-20b
LMArena Text: legal and government12801284Llama 3.3 70B Instruct
LMArena Text: business and finance12841243gpt-oss-20b
Neither model has a result on the other 6 boards here
Web designDirectional onlyPublic boards, shown as published, kept out of the index2 boards, none shared
Web design benchmarks for gpt-oss-20b and Llama 3.3 70B Instruct
Benchmarkgpt-oss-20bLlama 3.3 70B InstructLeader
Design Arena: websites852Not measuredOnly gpt-oss-20b measured
Design Arena: data viz933Not measuredOnly gpt-oss-20b measured
Neither model has a result on the other 7 boards here
GeneralLike-for-likePublic boards, shown as published, kept out of the index1010
General benchmarks for gpt-oss-20b and Llama 3.3 70B Instruct
Benchmarkgpt-oss-20bLlama 3.3 70B InstructLeader
IFBench65.1%47.1%gpt-oss-20b
Long context reasoning (AA-LCR)34.7%15.7%gpt-oss-20b
DTBench68.0%59.5%gpt-oss-20b
LMCA14.5%17.5%Llama 3.3 70B Instruct
Fiction.LiveBenchNot measured33.3%Only Llama 3.3 70B Instruct measured
LMArena Text12881274gpt-oss-20b
LMArena Text: instruction following12361242Llama 3.3 70B Instruct
LMArena Text: creative writing12011250Llama 3.3 70B Instruct
LMArena Text: English13031303Llama 3.3 70B Instruct
LMArena Text: non-English12691236gpt-oss-20b
LMArena Text: Chinese13151217gpt-oss-20b
LMArena Text: German12551251gpt-oss-20b
LMArena Text: Japanese12441150gpt-oss-20b
LMArena Text: Korean12361142gpt-oss-20b
LMArena Text: FrenchNot measured1282Only Llama 3.3 70B Instruct measured
LMArena Text: Spanish12671269Llama 3.3 70B Instruct
LMArena Text: Russian12781252gpt-oss-20b
LMArena Text: Polish12521259Llama 3.3 70B Instruct
LMArena Text: multi-turn12681280Llama 3.3 70B Instruct
LMArena Text: longer queries12501255Llama 3.3 70B Instruct
LMArena Text: writing and language12381255Llama 3.3 70B Instruct
LMArena Text: entertainment and media12161241Llama 3.3 70B Instruct
Neither model has a result on the other 24 boards here
Image generationNot comparablePublic boards, shown as published, kept out of the index0 boards, none shared
Image generation benchmarks for gpt-oss-20b and Llama 3.3 70B Instruct
Benchmarkgpt-oss-20bLlama 3.3 70B InstructLeader
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

gpt-oss-20b$0.0001Fits in one request
Llama 3.3 70B Instruct$0.0008Fits in one request

gpt-oss-20b costs less

Repository review

50K input + 3K output tokens

gpt-oss-20b$0.0019Fits in one request
Llama 3.3 70B Instruct$0.025Fits in one request

gpt-oss-20b costs less

Agent loop

200K input + 10K output tokens

gpt-oss-20b$0.0075Exceeds the context window
Llama 3.3 70B Instruct$0.097Exceeds the context window

gpt-oss-20b 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 gpt-oss-20b and Llama 3.3 70B Instruct
SpecificationOpenAI gpt-oss-20bMeta Llama 3.3 70B Instruct
MakerOpenAIMeta Llama
WeightsOpen weightsOpen weights
ReleasedAugust 5, 2025December 6, 2024
Context window131K131K
Input $/Mtok$0.03$0.45
Output $/Mtok$0.15$0.72
Coding indexNot benchmarkedNot benchmarked
Independent measurements behind it00
Cost per solved taskNot computableNot computable
SourceOpenRouter listingOpenRouter listing

Questions

Is gpt-oss-20b better than Llama 3.3 70B Instruct for coding?

Neither model has a coding score yet, so the two cannot be ranked on coding. Pick gpt-oss-20b for output tokens 4.8x cheaper ($0.15 vs $0.72 per million) and work measured by LMArena Text: legal and government or LMCA, where it leads.

Which is cheaper, gpt-oss-20b or Llama 3.3 70B Instruct?

gpt-oss-20b costs $0.03 per million input tokens and $0.15 per million output tokens; Llama 3.3 70B Instruct costs $0.45 and $0.72. Cost per solved task is known for neither of them.

Which has the bigger context window, gpt-oss-20b or Llama 3.3 70B Instruct?

Neither: both take 131K tokens.

How many benchmarks is the gpt-oss-20b vs Llama 3.3 70B Instruct comparison based on?

44 benchmarks have a result for both models as of October 22, 2026. gpt-oss-20b's coding index rests on 0 independent benchmarks and Llama 3.3 70B Instruct's on 0. Where only one model has a result, the page says so instead of naming a winner.

Compare next