Leaderboard | Coding

Show inactive models

Hide models that are no longer actively available on Yupp.

Turns

Filter model performance by the number of turns in a conversation.

Open license models

Filter the leaderboard to only show models that have an open license.

965

Llama 4 Scout

941

NVIDIA Llama 3.3 Nemotron Super 49B v1.5

935

Dobby Unhinged Llama 3.3 70B

933

Llama 3.3 70B

933

Llama 3.1 70B Instruct Turbo

919

Llama 3.3 Swallow 70B Instruct

904

Llama 3 8B

832

NVIDIA Llama 3.1 Nemotron Ultra 253B v1

623

Shisa V2 Llama 3.3 70B

595

DeepSeek-R1 Distill Llama 8B

Last updated about 1 month ago

Rank	Overall	Name	VIBE Score	Confidence Interval	Votes	Downvote %	Abort %	Speed	Latency	Context	Cost (Input)	Cost (Output)
1	167	Llama 4 Scout	965	±5	17.5K	5.3%	0.6%	88 tps	5.1s	131K	$0.18	$0.46
2	179	NVIDIA Llama 3.3 Nemotron Super 49B v1.5	941	±19	1.4K	6.9%	2.0%	50 tps	0.6s	131K	$0.09	$0.33
3	230	Dobby Unhinged Llama 3.3 70B	935	±19	860	2.8%	<0.1%	41 tps	0.4s	128K	$0.90	$0.90
4	189	Llama 3.3 70B	933	±10	3K	6.6%	0.3%	500 tps	0.5s	8K	$0.48	$0.66
5	245	Llama 3.1 70B Instruct Turbo	933	±11	4.1K	3.8%	<0.1%	110 tps	0.8s	128K	$0.88	$0.88
6	189	Llama 3.3 Swallow 70B Instruct	919	±8	3.5K	5.5%	1.4%	153 tps	1.3s	131K	$0.13	$0.39
7	201	Llama 3 8B	904	±10	3.2K	3.5%	6.0%	85 tps	0.7s	8K	$0.12	$0.16
8	324	NVIDIA Llama 3.1 Nemotron Ultra 253B v1	832	±16	2.2K	4.1%	<0.1%	40 tps	0.8s	128K	$0.30	$0.90
9	386	Shisa V2 Llama 3.3 70B	623	±24	585	9.3%	<0.1%	8 tps	2.0s	33K	$0.03	$0.09
10	390	DeepSeek-R1 Distill Llama 8B	595	±19	1.2K	5.6%	<0.1%	17 tps	N/A	32K	$0.04	$0.04