Fun-ASR-Nano ⭐ Flagship / default
-
-
Native Transformers support is merged upstream. Use the official -hf checkpoint, distinct from original, -vllm and GGUF artifacts. The pinned installation and inference guide explains why stable 5.16.1 is not sufficient.
-
Next-generation LLM-based ASR model. Combines SenseVoice's audio encoder with Qwen3-0.6B language model for superior context understanding. Supports vLLM acceleration for high-throughput batch inference and real-time streaming. The released
model.pt checkpoint does not provide reliable checkpoint-native character timestamps (
issue #106).
+
+
Start with released Transformers 5.17.0 and the official -hf checkpoint. CPU quickstart and batches · Try online · Notebook. Text transcription for zh/en/ja; original, -vllm and GGUF artifacts use different interfaces.
+
SenseVoice audio encoder plus Qwen3-0.6B decoder. Native Transformers produces text; service and edge paths use their own vLLM or GGUF artifacts. Timestamp and speaker capabilities depend on the exact checkpoint and pipeline: see the
model capability boundaries.
-vLLM accelerated
+vLLM runtime
zh/en/ja + Chinese dialects/accents
-Streaming
-LLM-quality
+Transformers
+Context hints
When to use
Best for high-throughput batch processing, real-time subtitles, and scenarios where LLM-quality context understanding improves output (e.g., proper nouns and code-switching). For reliable character-level timestamps, use Paraformer.
diff --git a/web-pages/product-site/legacy/models.html b/web-pages/product-site/legacy/models.html
index bd11a2d947..80ae538399 100644
--- a/web-pages/product-site/legacy/models.html
+++ b/web-pages/product-site/legacy/models.html
@@ -88,14 +88,14 @@
ASR 模型
Fun-ASR-Nano ⭐ 旗舰 / 默认推荐
-
-
原生 Transformers 路径已合入上游：使用 官方 -hf checkpoint，不要与原始权重、-vllm 或 GGUF 混用。固定版本的安装与推理指南说明 5.16.1 稳定版尚未包含的边界。
-
新一代 LLM 架构 ASR 模型。融合 SenseVoice 音频编码器与 Qwen3-0.6B 语言模型，具备更强的上下文理解能力。支持 vLLM 加速的高吞吐批量推理与实时流式服务。发布的
model.pt checkpoint 不提供可靠的原生字级时间戳（
issue #106）。
+
+
从正式版 Transformers 5.17.0 和 官方 -hf 权重开始。CPU 快速上手与批处理 · 在线体验 · Notebook。支持中英日文本转写；原始、-vllm 和 GGUF 权重的接口不同。
+
融合 SenseVoice 音频编码器与 Qwen3-0.6B 解码器。原生 Transformers 返回文本；服务与端侧分别使用对应的 vLLM、GGUF 权重。时间戳和说话人能力取决于具体 checkpoint 与流水线，见
模型能力边界。
-vLLM 加速
+vLLM 服务
中/英/日 + 中文方言/口音
-实时流式
-LLM 级别
+Transformers
+上下文提示
适用场景
适合高吞吐批处理、实时字幕，以及 LLM 级别上下文理解可提升输出质量的场景（如专有名词、代码混合语音）；需要可靠字级时间戳时请使用 Paraformer。
diff --git a/web-pages/product-site/tests/browser/native-transformers.spec.ts b/web-pages/product-site/tests/browser/native-transformers.spec.ts
index 0efedf89eb..e666ce3703 100644
--- a/web-pages/product-site/tests/browser/native-transformers.spec.ts
+++ b/web-pages/product-site/tests/browser/native-transformers.spec.ts
@@ -43,6 +43,8 @@ for (const prefix of ['', 'en/']) {
'href', new RegExp(`/docs/transformers_native${prefix ? '' : '_zh'}\\.md$`),
);
await expect(page.locator('.docs-article')).toContainText('torchaudio==2.10.0+cpu');
+ await expect(page.locator('.docs-article')).toContainText('transformers==5.17.0');
+ await expect(page.locator('.docs-article a[href*="fun_asr_nano_transformers.ipynb"]')).toBeVisible();
expect(await page.evaluate(() => document.documentElement.scrollWidth - document.documentElement.clientWidth)).toBeLessThanOrEqual(1);
await page.screenshot({ path: testInfo.outputPath('guide-top.png') });
await page.goto(`/${prefix}models.html`);
diff --git a/web-pages/product-site/tests/test_native_transformers_docs.py b/web-pages/product-site/tests/test_native_transformers_docs.py
index e2fda7b700..b09f44fce7 100644
--- a/web-pages/product-site/tests/test_native_transformers_docs.py
+++ b/web-pages/product-site/tests/test_native_transformers_docs.py
@@ -48,10 +48,12 @@ def test_catalogue_and_pages_export_own_both_native_guides():
def test_native_guide_pins_format_runtime_and_audio_dependency(suffix):
text = guide(suffix)
for expected in (MODEL, REVISION, HEAD, "torchaudio==2.10.0+cpu",
- "torch==2.10.0+cpu", "5.16.1", "2026-09-09",
+ "torch==2.10.0+cpu", "transformers==5.17.0", "2026-09-09",
"vllm", "GGUF", "46180"):
assert expected in text
- assert re.search(r"not.*stable|stable.*not|稳定版.*不|不.*稳定版", text, re.I)
+ assert "huggingface/transformers/archive/" not in text
+ assert "examples/transformers" in text
+ assert text.index("native-example: transcribe") < text.index("native-example: processor")
assert re.search(r"not.*(?:capacity|accuracy)|不.*(?:容量|准确率)", text, re.I)
@@ -77,7 +79,28 @@ def test_bilingual_recipes_are_identical_and_use_fixed_native_checkpoint(name):
assert "inference_mode" in en and "max_new_tokens" in en
assert "generated[:, inputs.input_ids.shape[1]:]" in en
assert "batch_decode" in en
- assert "16000" in en and "ndim" in en
+ assert "16000" in en and "example/en.mp3" in en
+ assert "272c57b82523ada6fd87095e955f8e29100979ab" in en
+
+
+@pytest.mark.parametrize("suffix", ["", "_zh", "_ja", "_ko"])
+def test_all_language_entry_points_reach_native_examples(suffix):
+ for name in [f"README{suffix}.md", f"docs/model_selection{suffix}.md",
+ f"docs/deployment_matrix{suffix}.md", f"examples/colab/README{suffix}.md"]:
+ text = (ROOT / name).read_text()
+ assert "5.17.0" in text, name
+ assert "5.16.1" not in text, name
+ text = (ROOT / f"README{suffix}.md").read_text()
+ assert "examples/transformers" in text
+ assert "fun_asr_nano_transformers.ipynb" in text
+
+
+def test_hf_catalogue_and_sphinx_reach_native_model():
+ text = (ROOT / "model_zoo/huggingface_models.md").read_text()
+ assert MODEL in text and "transformers_native.md" in text
+ text = (ROOT / "docs/index.rst").read_text()
+ assert " transformers_native\n" in text
+ assert " transformers_native_zh\n" in text
@pytest.mark.parametrize("name,heading", [
