Nova Pro

+ contribute

Nova Pro is a closed-weights language model from Amazon, released December 2024.

Catalogue identity notes

Arena entry 'amazon-nova-pro-v1.0' (text, vision). LMArena's leaderboard table (arena-leaderboard space @6f2eeeb) labels 'amazon-nova-pro-v1.0' as 'Amazon Nova Pro 1.0' (Amazon, Proprietary), linking https://docs.aws.amazon.com/nova/latest/userguide/what-is-nova.html. Released 2024-12: Bedrock model card: 'Model launch date: Dec 05, 2024'. Bedrock model id amazon.nova-pro-v1:0, 'Model lifecycle: Active' (Bedrock model card). First-generation Nova Pro; distinct from nova-2-pro.

49.5±1.2#352 of 719 LLMsfrom 17 benchmark results across 6 sources

$ how it compares

The LLMs ranked either side of Nova Pro, out of 828 with a score, then more from Amazon.

#	model	score	where	RAM	released
421	Llama 3 70B Instruct Meta	50±0	local	~24GB	2024-04
422	WizardLM-2 8x22B Microsoft	50±1	local	~91GB	2024-04
423	Nova Pro Amazon	50±1	api $1.40/M		2024-12
424	WizardCoder 33B V1.1 WizardLM	50±1	local	~22GB	2024-01
425	Mistral Small Instruct 22B (2409) Mistral AI	49±2	local	~15GB	2024-09
	more from Amazon
184	Nova 2 Pro (Preview) Amazon	64±2	api $3.78/M		2025-11
269	Nova 2 Lite Amazon	58±2	api $0.94/M		2025-10
339	Nova Premier Amazon	54±2	api $5.00/M		2025-04

see all LLMs · find the ones that fit your machine

$ what it's good at

agent tasks▁▁▃▆█▆▅▅▅▄▃▁ #212 of 300 · 14% of its score · 4 tests

Beats o1 (2024-12-17), o3-mini, DeepSeek R1 (Jan '25) here, though they rank higher overall. Trails SmolLM3 3B, Jamba 1.5 Mini, Granite 4.1 3B here, though they rank lower overall.

test	result	vs the others in its run	difficulty	counts	why
BFCL V4 (December 2025 archive)	25.0% ✎	▁▁▁▂██▅▅▆▂▂▂ beats 14/77	hard · avg 32%	▇▇ 5.3%	at its level
BFCL v3 +1	60.8% ✎	▁▁▁▁▁▃▅█▅▄▁▁ beats 83/114	medium · avg 52%	▇▇ 4.3%	blunt · at its level · several runs
Galileo Agent Leaderboard v1 (tool selection quality)	86.8% ✎	▃▃▃▃▆▅▆▅█▂▅▂ beats 17/31	easy · avg 79%	▇ 2.4%	blunt
Galileo Agent Leaderboard v2 (action completion)	29.0% ✎	▅▃·▃▃▅▅█▃█▅▅ beats 4/19	hard · avg 27%	▇ 2.0%	at its level

coding▁▁▁▃▆██▇▆▄▃▁ #263 of 458 * · 27% of its score · 2 tests

Beats Qwen3-VL 32B Instruct, Solar Open 100B (102B-A12B), GLM-4.7-Flash here, though they rank higher overall. Trails DeepSeek LLM 67B Chat, QwQ-32B-Preview, Phi-3 Medium 128K Instruct here, though they rank lower overall.

test	result	vs the others in its run	difficulty	counts	why
Aider code editing (133 Exercism Python exercises, two attempts)	54.1% ✎	▁·▁▂▄▁▆▄█▆▂▂ beats 19/62	medium · avg 54%	▇▇▇▇▇▇▇▇▇ 24%	at its level
BigCodeBench lab	22.3% ✎	▅····▃·█▃▅█▃ beats 2/12	hard · avg 29%	▇ 3.2%	at its level · small run

chat & writing▁▁▁▃▅▄▅█▆▆▇▃ #232 of 393 * · 12% of its score · 1 test

Beats Mistral Small 4 (119B-A6.5B), Ministral 3 14B Reasoning 2512, Qwen2.5 Coder 32B Instruct here, though they rank higher overall. Trails Flan-T5 XXL, LFM2.5 1.2B Instruct, Gemma 3 4B here, though they rank lower overall.

test	result	vs the others in its run	difficulty	counts	why
LMArena (text)	1290.1 ✎	▁▁▁▂▃▃▃▆▆▇█▃ beats 124/379	avg 1268	▇▇▇▇▇ 12%	blunt

math█··········█ #2 of 2 * · 13% of its score · 2 tests

test	result	vs the others in its run	difficulty	counts	why
AIME 2025 lab	5.3% ✎	▁·▁▃▁▂▅▆█▇▅▁ beats 4/133	hard · avg 16%	▇▇▇ 7.5%	sharp · small run
MATH-500 lab	76.6% ✎	▁▃▂▃▅▇▅▇█▅▂▁ beats 20/84	easy · avg 81%	▇▇ 5.6%	small run

vision & documents▁▁▃▃▁▃▄▂▆▇█▄ #127 of 147 · 17% of its score · 3 tests

Trails InternVL2-26B, Qwen3.5 2B, Pixtral 12B (2409) here, though they rank lower overall.

test	result	vs the others in its run	difficulty	counts	why
Arena vision (style controlled)	1018.7 ✎	▁▁▂▃▂▂▄▅▅▆██ beats 14/145	avg 1068	▇▇▇ 8.4%	blunt
CharXiv descriptive questions lab	70.5% ✎	▂▄▄▅▇▄█▇▇▂▁▂ beats 13/58	easy · avg 74%	▇▇ 4.4%	at its level · small run
CharXiv (reasoning questions) lab	40.6% ✎	▂▅▃▅▅▆▆▅▆█▆▂ beats 19/100	medium · avg 40%	▇▇ 4.3%	at its level · small run

other tests4 not in a task · 16% of its score

test	result	vs the others in its run	difficulty	counts	why
GPQA Diamond lab	50.0% ✎	▁▁▁▁▂▄▅▅███▄ beats 86/310	hard · avg 48%	▇▇▇▇ 9.5%	sharp · at its level · small run
MMLU lab	85.9% ✎	▁▁▁▁▁▂▃▅█▃▆▃ beats 161/183	easy · avg 82%	▇▇ 6.0%	at its level · small run
Vectara hallucination leaderboard, 2025 dataset (HHEM-2.3, factual consistency rate)	98.2% ✎	▁▁▂▃▄▄█▆▃▃▂▁ beats 124/140	very easy · avg 94%	▇ 0.8%	blunt · maxed out · noisy
Vectara hallucination leaderboard (HHEM-2.3, factual consistency rate)	94.9% ✎	▁▂▁▂▅█▂▃▄▃▁▁ beats 91/98	very easy · avg 90%	▇ 0.0%	blunt

$ where to run it

Through an API

api · Amazon Bedrock · $1.40/M · $0.8 in / $3.2 out per 1M tokens · 300k context · docs.aws.amazon.com ↗ · price sourceUSD per 1M tokens, Standard tier, geo cross-Region inference and in-Region, US East (N. Virginia). Amazon Nova Pro (amazon.nova-pro-v1:0). Observed 2026-09-28 (page price data $0.0008/$0.0032 per 1K tokens).

$ how the score was worked out · Amazon vendor-reported benchm… 42%, Aider code editing benchmark… 24%, LMArena leaderboard dataset 21% and 3 more

Each result implies a score on its own, after adjusting for how hard its test is and who ran it. The score is their weighted average, with a small pull toward the middle for thinly tested models. Each test's weight is its counts column above; here it's added up by who ran the tests. How it works

source	results	share of the score	licence
Amazon vendor-reported benchmark scores for its own Nova models (Amazon Science technical reports) · lab-reported	8	▇▇▇▇▇▇▇▇▇▇▇▇▇ 42%	Factual vendor-reported score (not licensed content)
Aider code editing benchmark runs (old edit leaderboard)	1	▇▇▇▇▇▇▇ 24%	Apache-2.0
LMArena leaderboard dataset	2	▇▇▇▇▇▇ 21%	CC BY 4.0
Berkeley Function Calling Leaderboard result archive	2	▇▇ 8%	Apache-2.0
Galileo Agent Leaderboard (v1 and v2 results)	2	▇ 4%	Apache-2.0
Vectara hallucination leaderboard (vectara/hallucination-leaderboard repository)	2	▇ 1%	Apache-2.0
prior · pull toward the middle		▇ 1%
weighted average		= 49.5