Model comparison

GPT-5.5 vs Llama-3.3-70B-Instruct

GPT-5.5 is the stronger model overall, scoring 63.4 to 30.6 on the Noometry Index. Llama-3.3-70B-Instruct costs 73× less per token, which makes it the better buy when GPT-5.5's lead doesn't matter for your workload.

Last verified October 10, 2026. 28 shared benchmarks.

GPT-5.5 OpenAI

63.4

Rank #9 Confirmed

Llama-3.3-70B-Instruct

Summary

They share 28 benchmarks with published results for both. GPT-5.5 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 math, where GPT-5.5 leads 81.7 to 15.3.
The biggest single-benchmark swing is OTIS Mock AIME 2024-2025: 100% for GPT-5.5 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 $5 / $30 for GPT-5.5.
GPT-5.5 accepts more context: 1.05M tokens versus 128K.
Llama-3.3-70B-Instruct has downloadable open weights; the other is API-only.

Side by side

GPT-5.5 and Llama-3.3-70B-Instruct specifications
	GPT-5.5	Llama-3.3-70B-Instruct
Provider	OpenAI	Meta
Noometry Index	63.4	30.6
Released	2026-04-23	2024-12-06
Weights	Proprietary	Open
Context window	1.05M	128K
Max output	128K	4K
Input $ / M tokens	$5	$0.10
Output $ / M tokens	$30	$0.32
Results tracked	71	43

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

Category by category

Coding GPT-5.5 leads

GPT-5.5: 58.2 (#17), Llama-3.3-70B-Instruct: 31.0 (#290)

Coding benchmarks
Benchmark	GPT-5.5	Llama-3.3-70B-Instruct
SciCode	56.1%	26%
WeirdML	84.9%	14.4%
LMArena Coding	1494	1268
SWE-bench Verified	80.6%	—
DeepSWE	67%	—
FrontierCode	43%	—
LMArena WebDev	1513	—
GSO	40.2%	—
BigCodeBench Instruct	—	46.9%
LiveBench Coding	—	36.6%
MirrorCode	10%	—
BigCodeBench Complete	—	57.5%
ALE-Bench	1,943	—

Agentic & Tool Use GPT-5.5 leads

GPT-5.5: 50.7 (#6), Llama-3.3-70B-Instruct: 25.8 (#105)

Agentic & Tool Use benchmarks
Benchmark	GPT-5.5	Llama-3.3-70B-Instruct
Terminal-Bench	84.7%	—
APEX-Agents	55.1%	—
Berkeley Function Calling Leaderboard	—	31.9%
OSWorld 2.0	13%	—
Remote Labor Index	6.3%	—
τ²-bench Banking	44.6%	—
DeepResearch Bench	54%	—
PostTrainBench	27.2%	—
BALROG	—	23%
ExploitBench	47.4%	—
GBAEval	53.2%	—
GDP.pdf	26%	—
LMArena Search	1242	—
Vending-Bench 2	7,524	—

Reasoning GPT-5.5 leads

GPT-5.5: 72.8 (#11), Llama-3.3-70B-Instruct: 14.1 (#327)

Reasoning benchmarks
Benchmark	GPT-5.5	Llama-3.3-70B-Instruct
SimpleBench	69%	19.9%
CritPt	27.1%	0%
LMArena Hard Prompts	1489	1257
DTBench	96%	59.5%
LMCA	54.3%	17.5%
Epoch Capabilities Index	159.1	127.33
ForecastBench	60.6	58.6
ARC-AGI-2	85%	—
Kagi LLM Benchmark	88.8%	—
NYT Connections (extended)	96.2%	—
ARC-AGI-1	95%	—
Chess Puzzles	54%	—
EBR-Bench	34.3%	—
LiveBench Reasoning	—	50.8%
Mystery Game Puzzles	56%	—
LiveBench Data Analysis	—	49.5%
Surface Evolver Bench	88.1%	—
Bench to the Future 3	0.14	—
LiveBench	—	50.2%

Math GPT-5.5 leads

GPT-5.5: 81.7 (#11), Llama-3.3-70B-Instruct: 15.3 (#298)

Math benchmarks
Benchmark	GPT-5.5	Llama-3.3-70B-Instruct
OTIS Mock AIME 2024-2025	100%	5.1%
LMArena Math	1486	1267
FrontierMath (Tiers 1-3)	85.3%	—
FrontierMath Tier 4	72.5%	—
MathArena Final-Answer Competitions	94.3%	—
ProofBench	50%	—
LiveBench Math	—	42.2%
MATH Level 5	—	41.6%
FrontierMath (Feb 2025 set)	51.7%	—
FrontierMath Erdős	0%	—
FrontierMath Tier 4 (v1)	35.4%	—

Knowledge GPT-5.5 leads

GPT-5.5: 64.4 (#17), Llama-3.3-70B-Instruct: 30.6 (#226)

Knowledge benchmarks
Benchmark	GPT-5.5	Llama-3.3-70B-Instruct
GPQA Diamond	94%	47.4%
Vectara Hallucination Rate	9.3%	4.1%
LMArena Expert	1508	1225
SimpleQA Verified	63%	—
Confabulations	—	22.8%
MMLU	—	86.3%

Multimodal Not comparable

GPT-5.5: 46.9 (#12), Llama-3.3-70B-Instruct: —

Multimodal benchmarks
Benchmark	GPT-5.5	Llama-3.3-70B-Instruct
LMArena Vision	1297	—
Blueprint-Bench 2	36.2%	—
Furniture Assembly	44.2%	—
LMArena Document	1486	—

Multilingual GPT-5.5 leads

GPT-5.5: 56.4 (#20), Llama-3.3-70B-Instruct: 39.9 (#220)

Multilingual benchmarks
Benchmark	GPT-5.5	Llama-3.3-70B-Instruct
LMArena Non-English	1467	1236
LMArena Chinese	1533	1217
LMArena French	1486	1281
LMArena German	1480	1251
LMArena Japanese	1498	1150
LMArena Korean	1460	1143
LMArena Russian	1473	1252
LMArena Spanish	1468	1270

Instruction Following GPT-5.5 leads

GPT-5.5: 77.5 (#18), Llama-3.3-70B-Instruct: 71.1 (#157)

Instruction Following benchmarks
Benchmark	GPT-5.5	Llama-3.3-70B-Instruct
LMArena Instruction Following	1479	1242
LiveBench Instruction Following	—	82.7%

Long Context GPT-5.5 leads

GPT-5.5: 48.3 (#12), Llama-3.3-70B-Instruct: 26.4 (#295)

Long Context benchmarks
Benchmark	GPT-5.5	Llama-3.3-70B-Instruct
LMArena Longer Query	1484	1256
Fiction.LiveBench	—	33.3%
CL-bench Life	22.2%	—

Writing & Preference GPT-5.5 leads

GPT-5.5: 72.7 (#13), Llama-3.3-70B-Instruct: 47.6 (#207)

Writing & Preference benchmarks
Benchmark	GPT-5.5	Llama-3.3-70B-Instruct
LMArena Text	1472	1274
LMArena Creative Writing	1455	1250
LMArena Multi-Turn	1476	1280
EQ-Bench Creative Writing	1844	—
EQ-Bench 4	1315	—
LiveBench Language	—	39.2%

Frequently asked questions

Is GPT-5.5 better than Llama-3.3-70B-Instruct?

Which is cheaper, GPT-5.5 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-5.5 lists at $5 and $30.

Is GPT-5.5 or Llama-3.3-70B-Instruct better for coding?

GPT-5.5 scores higher on coding benchmarks: 58.2 versus 31.0 in the Noometry coding category.

Which has the bigger context window?

GPT-5.5 does, with 1.05M tokens against 128K.

How many benchmarks do GPT-5.5 and Llama-3.3-70B-Instruct share?

28 benchmarks have published results for both models. GPT-5.5 has 71 scored results on Noometry and Llama-3.3-70B-Instruct has 43.