Instructions to use leoncca/Qwen3.8-Flash-Next-AWQ-g32 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Local Apps Settings

How to use leoncca/Qwen3.8-Flash-Next-AWQ-g32 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "leoncca/Qwen3.8-Flash-Next-AWQ-g32"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "leoncca/Qwen3.8-Flash-Next-AWQ-g32",
		"messages": [
			{
				"role": "user",
				"content": [
					{
						"type": "text",
						"text": "Describe this image in one sentence."
					},
					{
						"type": "image_url",
						"image_url": {
							"url": "https://cdn.britannica.com/61/93061-050-99147DCE/Statue-of-Liberty-Island-New-York-Bay.jpg"
						}
					}
				]
			}
		]
	}'

Use Docker

docker model run hf.co/leoncca/Qwen3.8-Flash-Next-AWQ-g32

SGLang

How to use leoncca/Qwen3.8-Flash-Next-AWQ-g32 with SGLang:

Install from pip and serve model

# Install SGLang from pip:
pip install sglang
# Start the SGLang server:
python3 -m sglang.launch_server \
    --model-path "leoncca/Qwen3.8-Flash-Next-AWQ-g32" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "leoncca/Qwen3.8-Flash-Next-AWQ-g32",
		"messages": [
			{
				"role": "user",
				"content": [
					{
						"type": "text",
						"text": "Describe this image in one sentence."
					},
					{
						"type": "image_url",
						"image_url": {
							"url": "https://cdn.britannica.com/61/93061-050-99147DCE/Statue-of-Liberty-Island-New-York-Bay.jpg"
						}
					}
				]
			}
		]
	}'

Use Docker images

docker run --gpus all \
    --shm-size 32g \
    -p 30000:30000 \
    -v ~/.cache/huggingface:/root/.cache/huggingface \
    --env "HF_TOKEN=<secret>" \
    --ipc=host \
    lmsysorg/sglang:latest \
    python3 -m sglang.launch_server \
        --model-path "leoncca/Qwen3.8-Flash-Next-AWQ-g32" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "leoncca/Qwen3.8-Flash-Next-AWQ-g32",
		"messages": [
			{
				"role": "user",
				"content": [
					{
						"type": "text",
						"text": "Describe this image in one sentence."
					},
					{
						"type": "image_url",
						"image_url": {
							"url": "https://cdn.britannica.com/61/93061-050-99147DCE/Statue-of-Liberty-Island-New-York-Bay.jpg"
						}
					}
				]
			}
		]
	}'

Docker Model Runner
How to use leoncca/Qwen3.8-Flash-Next-AWQ-g32 with Docker Model Runner:
```
docker model run hf.co/leoncca/Qwen3.8-Flash-Next-AWQ-g32
```

Qwen3.8-Flash-Next-AWQ-g32 / MODEL_PROVENANCE.json

leoncca

Add files using upload-large-folder tool

15e71db verified 5 days ago

Raw

History Blame Contribute Delete

6.51 kB

	{
	"artifact": {
	"name": "Qwen3.8-Flash-Next-AWQ-g32",
	"proposed_repository": "leoncca/Qwen3.8-Flash-Next-AWQ-g32",
	"proposed_tag": "v1-formal-r6-da62ba7b",
	"schema_version": 1
	},
	"sources": {
	"base_model": {
	"repository": "Qwen/Qwen3.8-Flash-Next",
	"revision": "f5d08274bafd880402bd16f5e3e6c514136ec06c"
	},
	"ple": {
	"repository": "Qwen/Qwen3.8-Flash-Next-FP8",
	"revision": "bcd9f01ddc9cff2316eb84281bebcd5b058bddce"
	},
	"license": {
	"filename": "LICENSE",
	"name": "Qwen Community License 1.0",
	"sha256": "a0dc422560841fd68e06d974907f8b4c709bca44a67daad2b528437bdf676c08"
	}
	},
	"quantization": {
	"method": "AWQ",
	"weight_bits": 4,
	"activation_dtype": "float16",
	"group_size": 32,
	"zero_point": true,
	"checkpoint_format": "gemm",
	"layout": "per-expert asymmetric routed-expert projections",
	"quantized_projection_names": [
	"gate_proj",
	"up_proj",
	"down_proj"
	],
	"protected": [
	"vision",
	"ple",
	"attention",
	"linear_attention",
	"router",
	"shared_expert",
	"embeddings",
	"lm_head",
	"hyper_connections",
	"mtp"
	],
	"quantizer": "gptqmodel:7.3.2",
	"calibration_record_count": 684,
	"calibration_corpus_sha256": "d2fbb16c5dcfaf26b713fd44d090cc8a3f2bbb31f1713f179e149f1f196b3dbb",
	"processor_audit_sha256": "1ca4205795e83f3cf341cb3d8eb10ef7169bb4934a4435878af7bef8507a8c05",
	"selective_bypass_selected_pairs": 93,
	"selective_bypass_target_manifest_sha256": "9d4c2372bb065d2f9aadce457d98b0314c3844e1989e8fc6e36ce2a7088770c9",
	"selective_bypass_source_report_sha256": "69d421dd1a57cf04ea55c533b20f042e9dc6a99e2b6f489c54aae3d2039b05a1",
	"expert_coverage": {
	"filename": "EXPERT_COVERAGE.json",
	"natural_unique_expert_ids": "512/512",
	"natural_layer_expert_pairs": "24568/24576",
	"natural_layer_expert_coverage_fraction": 0.9996744791666666,
	"fully_naturally_covered_layers": "45/48",
	"zero_natural_token_pairs": 8,
	"below_128_natural_token_pairs": 93,
	"selective_augmentation_pairs": 93,
	"runtime_zero_coverage_fallback_pairs": 0,
	"native_route_matrix_sha256": "6c987b6d2a1506978f38bc0b4fa167d205250ba8f0629f05adddc880a26828b1"
	},
	"group_size_rationale": {
	"expert_intermediate_width": 640,
	"validated_tensor_parallel_size": 4,
	"local_intermediate_width": 160,
	"g32_divides_local_width": true,
	"g128_divides_local_width": false,
	"g128_runtime_effective_group_size": 32,
	"native_g32_checkpoint_scale_and_zero_point_metadata_gib_approx": 8.79,
	"g128_checkpoint_scale_and_zero_point_metadata_gib_approx": 2.20,
	"native_g32_checkpoint_metadata_increase_gib_approx": 6.59,
	"tp4_runtime_metadata_increase_vs_expanded_g128": false,
	"decision": "Store native g32 so checkpoint groups align exactly with the TP4 expert partition and do not depend on load-time g128-to-g32 expansion.",
	"quality_claim_boundary": "Native g32 was quality-validated; no matched full-checkpoint g32-versus-g128 quality comparison was performed."
	}
	},
	"ple_splice": {
	"dtype": "float8_e4m3fn",
	"indexed_tensor_count": 129,
	"indexed_payload_bytes": 51200245762,
	"shard_count": 33,
	"scale_value": 0.00019931793212890625,
	"requantized": false
	},
	"checkpoint": {
	"index_filename": "model.safetensors.index.json",
	"base_awq_index_sha256": "da62ba7bfef57f8117f35fc78bf5daf14c0ab5eb1b8173b4d712c1f4d294819b",
	"index_sha256": "f066a0154a9101b359c3a4d4fa6a83fb12b7126a9cf0f19cb772611d08cc07ee",
	"indexed_tensor_count": 222771,
	"indexed_payload_bytes": 137042968666,
	"index_metadata_total_size": 137073585730,
	"index_metadata_total_size_contract": "AWQ shard file bytes plus indexed PLE tensor bytes plus complete QSA scale shard bytes",
	"physical_weight_shard_bytes": 138133206482,
	"physical_model_shard_bytes": 138133209122,
	"stored_tensor_count_in_all_selected_shards": 223827,
	"unindexed_tensor_count_in_reused_ple_shards": 1056,
	"awq_shard_count": 10,
	"ple_shard_count": 33,
	"qsa_kv_scale_shard_count": 1,
	"qsa_kv_scale_filename": "model-kvscales.safetensors",
	"qsa_kv_scale_tensor_count": 24,
	"qsa_kv_scale_file_sha256": "c89b08bacde7d869342190b2f95046a83af37080241ae00ccae3a3d6888852fd",
	"carrier_completion_sha256": "70fb852c73620edbf75858ab36c519005a5b4cec6597d2493c1355b4b901f077",
	"host_side_final_audit_sha256": "b854f56010a564d7db9636c9977154f10c4ac2280f852899802ad7145cf85475",
	"host_side_final_audit_status": "PASS"
	},
	"runtime_provenance": {
	"fp16_kv_quality_source_commit": "c9af1be6e11011e398e8f89808e62d43fa453031",
	"safetensors_index_assignment_commit": "1f8c110fb044a7ac3767e3b790d4efc936797dc9",
	"awq_tp_group_alignment_commit": "ad4bfbec41b4ffb20877e2895f07957685d6438b",
	"awq_tp_group_alignment_effect_for_g32_tp4": "inactive: effective group g32 and repeat factor 1",
	"qsa_e4m3_core_pr": "https://github.com/1CatAI/1Cat-vLLM/pull/447",
	"qsa_e4m3_core_commit": "58542ea69d43c6fc66242d218fbddbfdae7cd3b9",
	"qsa_e4m3_bitcast_pr": "https://github.com/1CatAI/1Cat-vLLM/pull/452",
	"qsa_e4m3_bitcast_commit": "6e42fb899767b4ac0769acaff9156314fb193d3a",
	"qsa_e4m3_scale_hoist_pr": "https://github.com/1CatAI/1Cat-vLLM/pull/453",
	"qsa_e4m3_scale_hoist_commit": "10a7bd0b36ff4684432b6e235013b7ad817a3534",
	"combined_performance_test_commit": "526570c756e456a867aab86ea3ffd345195a082b",
	"qwen4exp_multimodal_enablement_pr": "https://github.com/1CatAI/1Cat-vLLM/pull/458",
	"qwen4exp_multimodal_enablement_commit": "dd68b9e8f44d248ad7082b871243a66c5fc1696a"
	},
	"publication": {
	"internal_build_files_excluded": [
	"final-checkpoint-audit.json",
	"fp8-ple-splice.json",
	"quant_log.csv",
	"quantization-provenance.json"
	],
	"public_replacements": [
	"MODEL_PROVENANCE.json",
	"EXPERT_COVERAGE.json",
	"VALIDATION.json",
	"PUBLIC_RELEASE_AUDIT.json",
	"SHA256SUMS"
	],
	"config_sanitization": {
	"gpu_profile": "single-rtx-3090",
	"offload_to_disk_path_removed": true,
	"runtime_quantization_fields_changed": false
	},
	"weights_repacked": false,
	"model_index_changed": true,
	"model_index_change": "Added 24 checkpoint-bound QSA K/V scale assignments without changing AWQ or PLE weight payloads."
	}
	}