Skip to content
ModelCap

Amd

4 across 1 category board · 23 without a live endpoint · 42 equivalent routes · 8 superseded

Language ModelCap Index

4
Language rankModelGravityContextInput tokensOutput tokensReleased
#243HyLo Llama 14MLA14GDN 64K SFTOpen weights6.2—FreeFree10 days ago
#244HyLo Llama 6MLA22GDN 64K SFTOpen weights6.1—FreeFree10 days ago
#247HyLo Llama 4MLA12GDN 64K SFTOpen weights6.5—FreeFree10 days ago
#248HyLo Llama 8MLA8GDN 64K SFTOpen weights7.1—FreeFree10 days ago

No live endpoint

23
Board statusModelGravityContextInput tokensOutput tokensReleased
No live endpointHyLo Qwen 7MLA21GDN 64K SFTOpen weights
Language
6.2—FreeFree10 days ago
No live endpointHyLo Qwen 14MLA14GDN 64K SFTOpen weights
Language
6.3—FreeFree10 days ago
No live endpointHyLo Qwen 14MLA14M2 64K SFTOpen weights
Language
6.3—FreeFree10 days ago
No live endpointInstella MoE 16B A3B ThinkRestricted license
Language
11.5—FreeFree2 months ago
No live endpointzebra qwen3 7MLA21GDN noT SFT 1M combined fCEOpen weights
Language
5.6—FreeFree2 months ago
No live endpointtiny qwen3 moe w4a8Open weights
Language
10.4—FreeFree2 months ago
No live endpointPARD2 Llama 3.1 8BOpen weights
Language
5.2—FreeFree3 months ago
No live endpointPARD2 Qwen3 14BOpen weights
Language
6.9—FreeFree3 months ago
No live endpointPARD2 Qwen3 8BOpen weights
Language
6.2—FreeFree3 months ago
No live endpointDeepSeek OCROpen weights
Language
3.8—FreeFree7 months ago
No live endpointSAND MathScience DeepSeek Qwen32BRestricted license
Language
9.1—FreeFree9 months ago
No live endpointSAND Math Qwen2.5 32BRestricted license
Language
9.1—FreeFree9 months ago
No live endpointInstella 3B MathRestricted license
Language
8.6—FreeFreelast year
No live endpointPARD Qwen3 0.6BOpen weights
Language
8.8—FreeFreelast year
No live endpointPARD Llama 4 1BOpen weights
Language
4.6—FreeFreelast year
No live endpointInstella 3B Long InstructRestricted license
Language
8.3—FreeFreelast year
No live endpointPARD DeepSeek R1 Distill Qwen 1.5BOpen weights
Language
3.4—FreeFreelast year
No live endpointPARD Qwen2.5 0.5BOpen weights
Language
8.4—FreeFreelast year
No live endpointInstella 3B instructRestricted license
Language
8.1—FreeFreelast year
No live endpointInstella 3B Stage1Restricted license
Language
8.1—FreeFreelast year
No live endpointAMD OLMo 1BOpen weights
Language
6.3—FreeFreelast year
No live endpointAMD Llama 135m codeOpen weights
Language
3.3—FreeFree2 years ago
No live endpointAMD Llama 135mOpen weights
Language
11.4—FreeFree2 years ago

Reviewed equivalent routes

42
Board statusModelGravityContextInput tokensOutput tokensReleased
Equivalent routegpt oss 20b BF16 w4a16 llmcompressorOpen weights
Language
8.4—FreeFree3 days ago
Equivalent routegranite 4.0 h small w8a8 llmcompressorOpen weights
Language
7.4—FreeFree3 days ago
Equivalent routegranite 4.0 h small w4a16 llmcompressorOpen weights
Language
7.3—FreeFree3 days ago
Equivalent routegranite 4.0 h tiny w8a8 llmcompressorOpen weights
Language
7.7—FreeFree3 days ago
Equivalent routeQwen3.8 Flash Next Quark MXFP4 PLEFP8Restricted license
Language
12.4—FreeFree5 days ago
Equivalent routeQwen3.8 27B w8a8 llmcompressorOpen weights
Language
12.4—FreeFree6 days ago
Equivalent routeMuse Glimmer 30B w8a8 llmcompressorOpen weights
Language
12.4—FreeFree6 days ago
Equivalent routeInstella MoE 16B A3B SFT w8a8 llmcompressorRestricted license
Language
12.4—FreeFree6 days ago
Equivalent routeQwen3.5 397B A17B Quark MXFP4Open weights
Language
5.7—FreeFree11 days ago
Equivalent routeQwen3.8 Flash Next Quark MXFP4Restricted license
Language
12.0—FreeFree27 days ago
Equivalent routeQwen3.8 27B Quark AWQ MXFP4Open weights
Language
16.8—FreeFreelast month
Represented by Qwen3.8 27B Quark AWQ MXFP4Qwen3.8 27B Quark AWQ INT4 W4A16Open weights
Language
17.4—FreeFreelast month
Equivalent routeQwen3.8 27B Quark Qronos INT4 W4A16Open weights
Language
13.9—FreeFreelast month
Equivalent routeQwen3 VL 8B Instruct w4a16 llmcompressorOpen weights
Language
11.7—FreeFreelast month
Equivalent routeQwen3 VL 8B Instruct w8a8 llmcompressorOpen weights
Language
15.9—FreeFreelast month
Equivalent routeKimi K3 Quark MXFP4 AttnFP8Restricted license
Language
11.6—FreeFreelast month
Equivalent routeQwen3 VL 235B A22B Instruct MXFP4Gated access
Language
11.5—FreeFreelast month
Represented by Instella MoE 16B A3B ThinkInstella MoE 16B A3B SFTRestricted license
Language
11.5—FreeFree2 months ago
Represented by Instella MoE 16B A3B ThinkInstella MoE 16B A3B PretrainRestricted license
Language
11.5—FreeFree2 months ago
Represented by Instella MoE 16B A3B ThinkInstella MoE 16B A3B MidtrainRestricted license
Language
11.5—FreeFree2 months ago
Represented by Instella MoE 16B A3B ThinkInstella MoE 16B A3B DPORestricted license
Language
11.5—FreeFree2 months ago
Represented by Instella MoE 16B A3B ThinkInstella MoE 16B A3B BaseRestricted license
Language
11.5—FreeFree2 months ago
Equivalent routeQwen3.5 397B A17B MXFP4 AttnFP8 V2Open weights
Language
15.4—FreeFree2 months ago
Equivalent routeMiniMax M3 EAGLE3.1Open weights
Language
8.9—FreeFree2 months ago
Equivalent routeQwen3.5 397B A17B MoE MXFP4Open weights
Language
10.2—FreeFree3 months ago
Equivalent routeMiniMax M3 MXFP4 AttnFP8Restricted license
Language
10.9—FreeFree3 months ago
Equivalent routeMiniMax M3 MXFP4Restricted license
Language
10.9—FreeFree3 months ago
Equivalent routeQwen3 VL 8B Instruct w4a16 tao symgroup torchao v0.17.0Open weights
Language
3.7—FreeFree3 months ago
Equivalent routeQwen3 VL 8B Instruct w4a16 asym torchao v0.17.0Open weights
Language
12.2—FreeFree3 months ago
Represented by Qwen3.5 397B A17B MXFP4Qwen3.5 397B A17B NVFP4Open weights
Language
13.7—FreeFree3 months ago
Equivalent routeQwen3 VL 8B Instruct da8w8 torchao v0.17.0Open weights
Language
3.7—FreeFree3 months ago
Equivalent routegemma 4 31B it MXFP4Open weights
Language
8.4—FreeFree5 months ago
Equivalent routeQwen3.5 35B A3B MXFP4Open weights
Language
12.5—FreeFree6 months ago
Equivalent routeQwen3.5 397B A17B MXFP4Open weights
Language
14.7—FreeFree6 months ago
Equivalent routeKimi K2.5 W4A8Restricted license
Language
9.6—FreeFree7 months ago
Represented by Instella 3B MathInstella 3B Math SFTRestricted license
Language
8.6—FreeFreelast year
Represented by Instella 3B instructInstella 3B SFTRestricted license
Language
8.1—FreeFreelast year
Represented by Instella 3B instructInstella 3BRestricted license
Language
8.1—FreeFreelast year
Equivalent routeDeepSeek R1 Distill Llama 8B awq asym uint4 g128 lmheadLicense unverified
Language
8.0—FreeFreelast year
Represented by AMD OLMo 1BAMD OLMo 1B SFT DPOOpen weights
Language
4.0—FreeFreelast year
Represented by AMD OLMo 1BAMD OLMo 1B SFTOpen weights
Language
6.7—FreeFreelast year
Equivalent routeMeta Llama 3 8B fp8 quarkLicense unverified
Language
7.7—FreeFree2 years ago

Superseded versions

8
Board statusModelGravityContextInput tokensOutput tokensReleased
Superseded by HyLo Llama 6MLA22GDN 64K SFTHyLo Llama 6MLA22GDN 8K SFTOpen weights
Language
6.5—FreeFree10 days ago
Superseded by HyLo Llama 14MLA14GDN 64K SFTHyLo Llama 14MLA14GDN 8K SFTOpen weights
Language
6.2—FreeFree10 days ago
Superseded by HyLo Qwen 7MLA21GDN 64K SFTHyLo Qwen 7MLA21GDN 8K SFTOpen weights
Language
6.2—FreeFree10 days ago
Superseded by HyLo Qwen 14MLA14GDN 64K SFTHyLo Qwen 14MLA14GDN 8K SFTOpen weights
Language
6.3—FreeFree10 days ago
Superseded by HyLo Qwen 14MLA14M2 64K SFTHyLo Qwen 14MLA14M2 8K SFTOpen weights
Language
6.3—FreeFree10 days ago
Superseded by HyLo Llama 4MLA12GDN 64K SFTHyLo Llama 4MLA12GDN 8K SFTOpen weights
Language
6.3—FreeFree10 days ago
Superseded by HyLo Llama 8MLA8GDN 64K SFTHyLo Llama 8MLA8GDN 8K SFTOpen weights
Language
6.3—FreeFree10 days ago
Superseded by PARD Llama 4 1BPARD Llama 3.2 1BOpen weights
Language
10.0—FreeFreelast year

Amd models: common questions

How many AI models has Amd released?

ModelCap tracks 35 canonical Amd models as of 22 September 2026: 35 language models (4 with a public ModelCap Index position). Superseded versions and equivalent serving routes stay listed on this page, marked as such.

What is the best Amd model?

HyLo Llama 14MLA14GDN 64K SFT is Amd's highest-ranked model as of 22 September 2026, at ModelCap Index position #243 with a score of 4.1, followed by HyLo Llama 6MLA22GDN 64K SFT (#244). The table lists every ranked Amd model in Index order.

Are Amd models open-weight?

28 of 35 tracked Amd models publish downloadable weights (open weights). ModelCap labels weight availability from repository metadata, not whether the training stack is open source; the licence class is shown beside each row.

What is the newest Amd model?

HyLo Llama 6MLA22GDN 64K SFT, first listed on 11 September 2026 and currently ModelCap Index #244. New Amd models appear here as soon as they pass automatic identity checks; the launches page tracks releases across every publisher.

Explore further