Model rankings

Open models ranked by intelligence and speed, hosted on GPUs in the EU.

Try them for free Full catalog →

Intelligence leaderboard

Ranked by a composite of publicly reported benchmarks. Higher is better.

#	Model	Size	MMLU	GPQA	MATH	Code	Index
1	DeepSeek R1 0528 DeepSeek	—	91.4	81.0	97.0	92.0	90.4
2	GPT-OSS 120B Gpt-oss-120b	—	—	80.1	97.9	—	89.0
3	DeepSeek R1 DeepSeek	—	90.8	71.5	97.3	90.0	87.4
4	GLM 4.6 Z.AI	—	85.5	82.0	93.5	—	87.0
5	GLM 4.5 Z.AI	—	84.6	79.1	91.0	—	84.9
6	Qwen2.5 32B Qwen	32B	83.3	—	83.1	88.4	84.9
7	gpt oss 20b OpenAI	22B	—	71.5	96.0	—	83.8
8	Qwen3 235B A22B Qwen	235B	87.8	71.1	92.2	—	83.7
9	DeepSeek V3.1 DeepSeek	—	90.0	67.0	92.0	85.0	83.5
10	DeepSeek V3 (685B MoE) DeepSeek	685B	89.5	62.7	91.6	84.0	82.0
11	GLM 4.5 Air Z.AI	110B	81.0	75.0	89.0	—	81.7
12	Qwen 2.5 14B Qwen	14B	79.7	—	80.0	83.5	81.1
13	DeepSeek V3 DeepSeek	—	88.5	59.1	90.2	82.6	80.1
14	DeepSeek R1 Distill Llama 70B DeepSeek	70B	—	65.2	94.5	—	79.9
15	Qwen3 32B Qwen	32B	82.4	68.4	88.5	—	79.8
16	DeepSeek R1 Distill 32B DeepSeek	32B	—	62.1	94.3	—	78.2
17	Qwen3 14B Qwen	14B	80.7	65.8	87.7	—	78.1
18	Qwen3 30B A3B Qwen	30B	79.5	65.8	86.6	—	77.3
19	DeepSeek R1 Distill 14B DeepSeek	14B	—	59.1	93.9	—	76.5
20	Qwen 2.5 72B Qwen	72B	86.1	49.0	83.1	86.6	76.2
21	Phi 4 (14B) Microsoft	14B	84.8	56.1	80.4	82.6	76.0
22	Llama 3.3 70B Meta	70B	86.0	50.5	77.0	88.4	75.5
23	Llama 3.1 405B Instruct Meta	405B	87.3	50.7	73.8	89.0	75.2
24	Qwen 2.5 Coder 32B Qwen	32B	—	—	57.2	92.7	75.0
25	Qwen3 8B Qwen	8B	76.9	62.0	85.0	—	74.6
26	Qwen 2.5 Coder 7B Qwen	7B	—	—	57.2	88.4	72.8
27	DeepSeek R1 Distill Qwen 7B DeepSeek	7.6B	—	49.1	92.8	—	71.0
28	Llama 3.1 70B Meta	70B	83.6	41.7	68.0	80.5	68.5
29	Qwen2.5 7B Qwen	7B	74.2	36.4	75.5	84.8	67.7
30	Phi 4 Mini (3.8B) Microsoft	3.8B	67.3	—	64.0	—	65.7
31	Gemma 3 27B Google	27B	—	42.4	89.0	—	65.7
32	gemma 3 12b Google	12B	—	34.9	83.8	—	59.4
33	gemma 2 27b it Google	27B	75.2	—	42.3	—	58.8
34	Phi 3.5 mini instruct Microsoft	3.8B	69.0	—	48.5	—	58.8
35	Llama 3.1 8B Meta	8B	69.4	30.4	51.9	72.6	56.1
36	Llama 3.2 3B Meta	3B	63.4	—	48.0	—	55.7
37	gemma 2 9b it Google	9.2B	71.3	—	36.6	—	54.0
38	Gemma 3 4B Google	4B	—	30.8	75.6	—	53.2
39	Mistral Nemo 12BEU Mistral	12B	68.0	—	38.0	—	53.0
40	Llama 3.2 1B Meta	1B	49.3	—	30.6	—	40.0
41	Mistral 7B Instruct v0.2EU Mistral	7.2B	60.1	—	11.2	30.5	33.9

Benchmarks are publicly reported figures from model authors and open leaderboards (last updated 2026-06). Methodology varies per model, so the index is an ordering aid, not a controlled benchmark.

Speed leaderboard

Ranked by the real time-to-respond we measured on our own EU GPUs. Lower is better.

#	Model	Size	Response
1	codegemma 2b Google	2.5B	66 ms
2	gemma 2b it Google	2.5B	67 ms
3	Llama 3.2 3B Instruct Meta	3.2B	76 ms
4	Phi 3 mini 128k instruct Microsoft	3.8B	81 ms
5	Apertus v1.1 1.5B InstructEU Swiss-ai	1.5B	81 ms
6	Flor 6.3B Instruct 4096EU BSC-LT	6.2B	81 ms
7	Phi 3.5 mini instruct Microsoft	3.8B	82 ms
8	granite vision 3.1 2b preview Ibm-granite	3B	83 ms
9	Llama 3.2 1B Instruct Meta	1.2B	87 ms
10	Meta Llama 3 8B Instruct Meta	8B	89 ms
11	Qwen3 VL 2B Instruct Qwen	2.1B	93 ms
12	Phi 3.5 vision instruct Microsoft	4.1B	93 ms
13	Flor 6.3B InstructEU BSC-LT	6.2B	94 ms
14	Qwen3.5 0.8B Qwen	0.9B	96 ms
15	Qwen3.5 2B Qwen	2B	96 ms
16	phi 4 Microsoft	15B	98 ms
17	Qwen2.5 14B Instruct Qwen	15B	98 ms
18	txgemma 2b predict Google	2.6B	99 ms
19	Qwen2.5 VL 7B Surg CholecT50 NVIDIA	8.3B	101 ms
20	salamandra 7b instruct tools 16kEU BSC-LT	7.8B	102 ms
21	Phi 4 mini instruct Microsoft	3.8B	103 ms
22	salamandra 7b instruct 2606EU BSC-LT	7.8B	103 ms
23	X Reasoner 7B Microsoft	8.3B	106 ms
24	Llama Poro 2 8B SFTEU LumiOpen	8B	109 ms
25	EuroLLM 22B Instruct PreviewEU Utter-project	23B	109 ms

Response times were measured at our last verification of each model and vary with load and cold starts.

Host. Route. Ship.

No credit card required. Pay as you go, cancel anytime.

Start Hosting Free Today