Qwen3 14B

Tool callingStructured outputReasoning

Hugging Face Launch post Website Docs

Context: 41K
Released: 1y ago
Parameters: 14B
Input / 1M tokens: $0.12
Output / 1M tokens: $0.24

Sources:OpenRouter

Benchmarks

DeepSWE is the primary coding benchmark where an exact model evaluation exists. Other sourced evaluations appear below; missing scores are not zero.

Not evaluated in DeepSWE 1.1 as of 2026-09-22. Results are only shown for exact model versions in the official dataset. Official leaderboard.

Other published benchmarks

Evaluation	Reported score and source
GPQA diamond	63.8% Epoch AI · qwen3-14b Epoch AI evaluation
GPQA diamond	53.4% Epoch AI · qwen3-14b_none Reasoning effort: none Epoch AI evaluation
OTIS Mock AIME 2024-2025	66.4% Epoch AI · qwen3-14b Epoch AI evaluation
OTIS Mock AIME 2024-2025	25.8% Epoch AI · qwen3-14b_none Reasoning effort: none Epoch AI evaluation

Benchmark compilation: Epoch AI, Capabilities & benchmarking (CC BY attribution). Scores are converted to percentages and model names matched to the Vertical Tabs catalog. Individual evaluations are credited to their original sources. External results may carry separate licences and evaluation configurations; do not interpret scores from different harnesses as directly comparable.

Specifications

Developer: Qwen
Model family: Qwen
Type: Language model
Predecessor: —
Successor: —
Released: April 29, 2025
Model ID: qwen/qwen3-14b
Openness: Open weights
License: —
Parameters: 14B
Input modalities: text
Output modalities: text
Modality: text->text
Native context: 256K
Tokenizer: Qwen3
Max output tokens: 16K
Knowledge cutoff: 2025-03-31
Hugging Face: Qwen/Qwen3-14B
Source freshness: Endpoints synced Oct 15, 2026, 8:17 AM UTC

Open-weight builds

Ollama variants

Every published Ollama tag matched to this model, including parameter editions, instruction/base variants, and quantized builds.

41 builds

Variant	Parameters	Type	Quantization	Size	Context	Input
qwen3:0.6b-fp16	0.6B	—	FP16	1.5 GB	40K	text
qwen3:0.6b	0.6B	—	Q4_K_M	523 MB	40K	text
qwen3:0.6b-q8_0	0.6B	—	Q8_0	832 MB	40K	text
qwen3:1.7b-fp16	1.7B	—	FP16	4.1 GB	40K	text
qwen3:1.7b	1.7B	—	Q4_K_M	1.4 GB	40K	text
qwen3:1.7b-q8_0	1.7B	—	Q8_0	2.2 GB	40K	text
qwen3:4b-instruct-2507-fp16	4B	instruct	FP16	8.1 GB	256K	text
qwen3:4b-instruct	4B	instruct	Q4_K_M	2.5 GB	256K	text
qwen3:4b-instruct-2507-q8_0	4B	instruct	Q8_0	4.3 GB	256K	text
qwen3:4b-thinking-2507-fp16	4B	thinking	FP16	8.1 GB	256K	text
qwen3:4b	4B	thinking	Q4_K_M	2.5 GB	256K	text
qwen3:4b-thinking-2507-q8_0	4B	thinking	Q8_0	4.3 GB	256K	text
qwen3:4b-fp16	4B	—	FP16	8.1 GB	40K	text
qwen3:4b-q4_K_M	4B	—	Q4_K_M	2.6 GB	40K	text
qwen3:4b-q8_0	4B	—	Q8_0	4.4 GB	40K	text
qwen3:8b-fp16	8B	—	FP16	16 GB	40K	text
qwen3:latest	8B	—	Q4_K_M	5.2 GB	40K	text
qwen3:8b-q8_0	8B	—	Q8_0	8.9 GB	40K	text
qwen3:14b-fp16	14B	—	FP16	30 GB	40K	text
qwen3:14b	14B	—	Q4_K_M	9.3 GB	40K	text
qwen3:14b-q8_0	14B	—	Q8_0	16 GB	40K	text
qwen3:30b-a3b-instruct-2507-fp16	30B	instruct	FP16	61 GB	256K	text
qwen3:30b-a3b-instruct-2507-q8_0	30B	instruct	Q8_0	32 GB	256K	text
qwen3:30b-a3b-instruct-2507-q4_K_M	30B	instruct	—	19 GB	256K	text
qwen3:30b-a3b-thinking-2507-fp16	30B	thinking	FP16	61 GB	256K	text
qwen3:30b-a3b-thinking-2507-q8_0	30B	thinking	Q8_0	32 GB	256K	text
qwen3:30b	30B	thinking	—	19 GB	256K	text
qwen3:30b-a3b-fp16	30B	—	FP16	61 GB	40K	text
qwen3:30b-a3b-q4_K_M	30B	—	Q4_K_M	19 GB	40K	text
qwen3:30b-a3b-q8_0	30B	—	Q8_0	33 GB	40K	text
qwen3:32b-fp16	32B	—	FP16	66 GB	40K	text
qwen3:32b	32B	—	Q4_K_M	20 GB	40K	text
qwen3:32b-q8_0	32B	—	Q8_0	35 GB	40K	text
qwen3:235b-a22b-instruct-2507-q8_0	235B	instruct	Q8_0	250 GB	256K	text
qwen3:235b-a22b-instruct-2507-q4_K_M	235B	instruct	—	142 GB	256K	text
qwen3:235b-a22b-thinking-2507-fp16	235B	thinking	FP16	470 GB	256K	text
qwen3:235b-a22b-thinking-2507-q8_0	235B	thinking	Q8_0	250 GB	256K	text
qwen3:235b	235B	thinking	—	142 GB	256K	text
qwen3:235b-a22b-fp16	235B	—	FP16	470 GB	40K	text
qwen3:235b-a22b-q4_K_M	235B	—	Q4_K_M	142 GB	40K	text
qwen3:235b-a22b-q8_0	235B	—	Q8_0	250 GB	40K	text

Infrastructure

Provider availability

All providers

Provider	Endpoint	Context	Input /M tokens	Output /M tokens	Cached input /M tokens	Cache write /M tokens	Uptime 1d	Data handling
NextBit	nextbit/int4	41K	$0.10	$0.22	—	—	99.92%	ZDR: Yes Training: Unknown Prompt logging: Unknown Retention: Unknown ZDR source Privacy policy Checked 2026-10-15
DeepInfra	deepinfra/fp8	41K	$0.12	$0.24	—	—	98.21%	ZDR: Yes Training: Unknown Prompt logging: Unknown Retention: Unknown ZDR source Privacy policy Checked 2026-10-15

Provider data:OpenRouter

Deployment graph

Can I run this?

Vertical Tabs only shows compatibility when it has structured evidence tying this exact model to hardware, runtime, quantization, memory use, and observed performance.

Browse hardware