What are Llama 2-7B's benchmark scores?

As of 25 Oct 2026, Llama 2-7B scores 33.8 ±5.3 on the BenchLeader Index, strongest in composite (36) and weakest in coding (18). That rests on 5 independent benchmark results, and every score on this page links to its publisher.

How does Llama 2-7B rank?

Llama 2-7B is #775 of 785 ranked configurations on 25 Oct 2026. 100 other configurations score within its margin, so the exact position matters less than the band. The leader is Claude Opus 5.5 (max effort) at 72.6.

How much does Llama 2-7B cost per 1M tokens?

As of 25 Oct 2026, Llama 2-7B costs $0.050 per 1M input tokens and $0.250 per 1M output tokens, $0.100 blended (three input tokens per output token).

When was Llama 2-7B released?

Meta released Llama 2-7B on 18 Jul 2023.

Meta Open weights: meta-llama/Llama-2-7b Compare with another model

Llama 2-7B

Llama 2-7B is a Meta open-weights model, released 18 Jul 2023. Its best configuration ranks #775 of 785 on the BenchLeader Index at 33.8 ±5.3, in the lower half. The ± is the point: 100 other configurations score within that range, so they and this one cannot be told apart on quality alone — price and speed are what separate them. It scores highest in composite (36) and lowest in coding (18). At $0.100 per million tokens blended it is among the cheapest fifth of ranked models. Last measured 22 Oct 2026.

Blended price: $0.100/M; $0.050 in · $0.250 out
Output speed: –
First answer: –
Context: 4k
Full answer: –
Cost per run: –
Released: 18 Jul 2023

How it scores by categoryDashed line = average model (50). One step of 15 = one standard deviation.

Overall index34
Reasoning21
Coding18
Maths23
Human preference22
Composite36

Versions

Meta has shipped 2 models under this name. Each is ranked on its own results; a newer version often has fewer results so far, which holds its index nearer the average until more arrive.

Model	Released	Index	Rank
Llama 2-7Bthis page	18 Jul 2023	33.8	#775
LLaMA-7B	24 Feb 2023	–	–

How its index has moved

19 Sept 2026 to 19 Oct 2026

The index is recomputed from scratch every day, so a line moves when a new benchmark result lands, when a publisher revises a score, or when the models it is normalised against change. Early movement usually means the score is still settling.

Benchmark results

See all LLM benchmarks. One column per reasoning effort. Rank is among every configuration of every model on that benchmark. Hover a score for the run it came from.

Reasoning

Benchmark	not stated	Source
Arena Hard Prompts	1097#358	Arena (LMArena)
GPQA Diamond (AA)not in index	22.7%#553	Artificial Analysis
Humanity's Last Exam (AA)not in index	5.3%#433	Artificial Analysis
HellaSwagnot in index	77.2%#33	Epoch AI Benchmarking Hub
BIG-Bench Hardnot in index	39.2%#31	Epoch AI Benchmarking Hub
WinoGrandenot in index	69.2%#47	Epoch AI Benchmarking Hub
ARC (AI2 Challenge)not in index	45.9%#52	Epoch AI Benchmarking Hub
PIQAnot in index	78.8%#36	Epoch AI Benchmarking Hub

Coding

Benchmark	not stated	Source	Trend
Arena Coding	1103#356	Arena (LMArena)

Maths

Benchmark	not stated	Source	Trend
Arena Maths	1086#341	Arena (LMArena)
GSM8Knot in index	16.7%#63	Epoch AI Benchmarking Hub

Knowledge

Benchmark	not stated	Source
MMLUnot in index	45.8%#97	Epoch AI Benchmarking Hub
TriviaQAnot in index	73.7%#24	Epoch AI Benchmarking Hub
OpenBookQAnot in index	58.6%#15	Epoch AI Benchmarking Hub
LAMBADAnot in index	73.3%#14	Epoch AI Benchmarking Hub

Instruction following

Benchmark	not stated	Source	Trend
Arena Instruction Followingnot in index	1070#358	Arena (LMArena)

Human preference

Benchmark	not stated	Source
Arena Text	1108#357	Arena (LMArena)
Arena Creative Writingnot in index	1072#357	Arena (LMArena)
Arena Multi-turnnot in index	1078#346	Arena (LMArena)
Arena Longer Queriesnot in index	1078#349	Arena (LMArena)

Multimodal

Benchmark	not stated	Source	Trend
ScienceQAnot in index	43.1%#22	Epoch AI Benchmarking Hub

Composite

Benchmark	not stated	Source	Trend
Epoch Capabilities Indexnot in index	98.6#222	Epoch AI Benchmarking Hub
AA Intelligence Index v4.3.2	5.7#563	Artificial Analysis

What a task costs

Estimates from list price, output speed and time to first answer for the best configuration. “With caching” assumes three-quarters of the input is served from the prompt cache. Reasoning tokens are not modelled.

Workload	Tokens in / out	Cost	With caching	Time
Chat reply	400 / 300	$0.0001	–	–
Summarize a 30-page report	12,000 / 600	$0.0008	–	–
Code edit	6,000 / 1,500	$0.0007	–	–
Agentic coding session	60,000 / 4,000	$0.0040	–	–
Structured extraction	2,000 / 200	$0.0001	–	–

New this month

Questions people ask

What are Llama 2-7B's benchmark scores?: As of 25 Oct 2026, Llama 2-7B scores 33.8 ±5.3 on the BenchLeader Index, strongest in composite (36) and weakest in coding (18). That rests on 5 independent benchmark results, and every score on this page links to its publisher.
How does Llama 2-7B rank?: Llama 2-7B is #775 of 785 ranked configurations on 25 Oct 2026. 100 other configurations score within its margin, so the exact position matters less than the band. The leader is Claude Opus 5.5 (max effort) at 72.6.
How much does Llama 2-7B cost per 1M tokens?: As of 25 Oct 2026, Llama 2-7B costs $0.050 per 1M input tokens and $0.250 per 1M output tokens, $0.100 blended (three input tokens per output token).
When was Llama 2-7B released?: Meta released Llama 2-7B on 18 Jul 2023.

Data as of 25 Oct 2026. Compare with another model.

Cite as: BenchLeader, “Llama 2-7B: benchmarks, pricing, speed and rank”, data as of 25 Oct 2026, https://www.benchleader.com/models/llama-2-7b.