Noometry

Model comparison

GPT-5 vs Llama-3.3-70B-Instruct

GPT-5 is the stronger model overall, scoring 50.9 to 30.6 on the Noometry Index. Llama-3.3-70B-Instruct costs 22× less per token, which makes it the better buy when GPT-5's lead doesn't matter for your workload.

Last verified . 32 shared benchmarks.

GPT-5 OpenAI

50.9

Rank #45 Confirmed

Llama-3.3-70B-Instruct Meta

30.6

Rank #291 Confirmed

Summary

  • They share 32 benchmarks with published results for both. GPT-5 scores higher in 9 categories and Llama-3.3-70B-Instruct in 0 categories; 9 gaps are clear of the uncertainty.
  • The widest gap is in long context, where GPT-5 leads 69.5 to 26.4.
  • The biggest single-benchmark swing is OTIS Mock AIME 2024-2025: 91.4% for GPT-5 and 5.1% for Llama-3.3-70B-Instruct.
  • Llama-3.3-70B-Instruct is cheaper at $0.10 / $0.32 per million input/output tokens, against $1.25 / $10 for GPT-5.
  • GPT-5 accepts more context: 400K tokens versus 128K.
  • Llama-3.3-70B-Instruct has downloadable open weights; the other is API-only.

Side by side

GPT-5Llama-3.3-70B-Instruct
ProviderOpenAIMeta
Noometry Index50.930.6
Released2025-08-072024-12-06
WeightsProprietaryOpen
Context window400K128K
Max output128K4K
Input $ / M tokens$1.25$0.10
Output $ / M tokens$10$0.32
Results tracked6943

Sponsored placements are available on pages like this one. Advertise on Noometry

Category by category

Coding GPT-5 leads

GPT-5: 50.3 (#47), Llama-3.3-70B-Instruct: 31.0 (#290)

BenchmarkGPT-5Llama-3.3-70B-Instruct
SciCode42.9%26%
WeirdML60.7%14.4%
LMArena Coding14361268
SWE-bench Verified73.6%—
SWE-bench Verified (bash only)65%—
Aider Polyglot88%—
LMArena WebDev1418—
GSO6.9%—
BigCodeBench Instruct—46.9%
LiveBench Coding—36.6%
BigCodeBench Complete—57.5%
ALE-Bench1,162—
AlgoTune1.67—

Agentic & Tool Use GPT-5 leads

GPT-5: 33.1 (#56), Llama-3.3-70B-Instruct: 25.8 (#105)

BenchmarkGPT-5Llama-3.3-70B-Instruct
BALROG32.8%23%
Terminal-Bench49.6%—
Berkeley Function Calling Leaderboard—31.9%
GDPval34.8%—
Remote Labor Index1.7%—
DeepResearch Bench49.6%—
LMArena Search1133—
METR Time Horizons69.6%—

Reasoning GPT-5 leads

GPT-5: 38.3 (#64), Llama-3.3-70B-Instruct: 14.1 (#327)

BenchmarkGPT-5Llama-3.3-70B-Instruct
SimpleBench56.7%19.9%
CritPt12.6%0%
LMArena Hard Prompts14161257
DTBench90.7%59.5%
LMCA40%17.5%
Epoch Capabilities Index150127.33
ForecastBench61.458.6
ARC-AGI-29.9%—
Kagi LLM Benchmark72.7%—
ARC-AGI-165.7%—
Chess Puzzles37%—
EnigmaEval10.5%—
EBR-Bench12.7%—
LiveBench Reasoning—50.8%
Mystery Game Puzzles23%—
LiveBench Data Analysis—49.5%
LiveBench—50.2%

Math GPT-5 leads

GPT-5: 55.0 (#44), Llama-3.3-70B-Instruct: 15.3 (#298)

BenchmarkGPT-5Llama-3.3-70B-Instruct
OTIS Mock AIME 2024-202591.4%5.1%
LMArena Math14071267
MATH Level 598.1%41.6%
FrontierMath (Tiers 1-3)55.4%—
FrontierMath Tier 422%—
ProofBench18%—
Omni-MATH64.7%—
LiveBench Math—42.2%
FrontierMath (Feb 2025 set)32.4%—
FrontierMath Tier 4 (v1)12.5%—

Knowledge GPT-5 leads

GPT-5: 56.6 (#43), Llama-3.3-70B-Instruct: 30.6 (#226)

BenchmarkGPT-5Llama-3.3-70B-Instruct
GPQA Diamond86.2%47.4%
Confabulations10.3%22.8%
Vectara Hallucination Rate14.7%4.1%
LMArena Expert14191225
Humanity's Last Exam25.3%—
SimpleQA Verified50.1%—
MMLU-Pro86.3%—
GPQA (HELM)79.2%—
MMLU—86.3%

Multimodal Not comparable

GPT-5: 46.8 (#13), Llama-3.3-70B-Instruct: —

BenchmarkGPT-5Llama-3.3-70B-Instruct
LMArena Vision1232—
GeoBench81%—
VPCT66%—

Multilingual GPT-5 leads

GPT-5: 51.4 (#110), Llama-3.3-70B-Instruct: 39.9 (#220)

BenchmarkGPT-5Llama-3.3-70B-Instruct
LMArena Non-English13971236
LMArena Chinese14221217
LMArena French14101281
LMArena German14161251
LMArena Japanese14091150
LMArena Korean13601143
LMArena Russian14061252
LMArena Spanish13991270

Instruction Following GPT-5 leads

GPT-5: 73.8 (#113), Llama-3.3-70B-Instruct: 71.1 (#157)

BenchmarkGPT-5Llama-3.3-70B-Instruct
LMArena Instruction Following13881242
LiveBench Instruction Following—82.7%
IFEval87.5%—

Long Context GPT-5 leads

GPT-5: 69.5 (#2), Llama-3.3-70B-Instruct: 26.4 (#295)

BenchmarkGPT-5Llama-3.3-70B-Instruct
Fiction.LiveBench97.2%33.3%
LMArena Longer Query13991256

Writing & Preference GPT-5 leads

GPT-5: 63.4 (#65), Llama-3.3-70B-Instruct: 47.6 (#207)

BenchmarkGPT-5Llama-3.3-70B-Instruct
LMArena Text14061274
LMArena Creative Writing13651250
LMArena Multi-Turn14261280
Short-Story Creative Writing86%—
EQ-Bench Creative Writing1627—
WildBench85.7%—
LiveBench Language—39.2%

Frequently asked questions

Is GPT-5 better than Llama-3.3-70B-Instruct?

GPT-5 is the stronger model overall, scoring 50.9 to 30.6 on the Noometry Index. Llama-3.3-70B-Instruct costs 22× less per token, which makes it the better buy when GPT-5's lead doesn't matter for your workload.

Which is cheaper, GPT-5 or Llama-3.3-70B-Instruct?

Llama-3.3-70B-Instruct is cheaper. It lists at $0.10 per million input tokens and $0.32 per million output tokens; GPT-5 lists at $1.25 and $10.

Is GPT-5 or Llama-3.3-70B-Instruct better for coding?

GPT-5 scores higher on coding benchmarks: 50.3 versus 31.0 in the Noometry coding category.

Which has the bigger context window?

GPT-5 does, with 400K tokens against 128K.

How many benchmarks do GPT-5 and Llama-3.3-70B-Instruct share?

32 benchmarks have published results for both models. GPT-5 has 69 scored results on Noometry and Llama-3.3-70B-Instruct has 43.

Related comparisons

Go deeper