Loading video...

Video Failed to Load

Go Home

Which local models can actually handle tool calling? I built a framework to find out. 15 scenarios. 12 tools. Mocked responses. Temperature 0. No cherry-picking. Tested every Qwen3.5 size from 0.8B to 397B, and since some of you asked after the distillation tests: yes, I included Jackrong's Qwen3.5-27B-Claude-4.6-Opus-Reasoning-Distilled too....

stevibe's profile picture

stevibe

27,265 subscribers

428,501 views • 3 months ago •via X (Twitter)

EducationNews & PoliticsScience & Technology
LIVE

Anya Rossi• Live Now

Private livecam show

0 Comments

No comments available

Comments from the original post will appear here

Related Videos

🥊 Qwen3.6 35B A3B vs Qwen3.6 27B Made them fight on the same prompt 🌊 Ocean Waves, a canvas challenge 35B took 39.7s (at 142tok/s) 27B took 111.6s (at 50tok/s) Both models are really good. The fact that the 35B produced such a strong result in under 40 seconds is seriously impressive. But when you look at 27B's output... it's actually so much better Added clouds, beautiful foam particles and splashing effects, a perfectly straight horizon, more realistic blinking stars, and a nice sun reflection on the water 🏆 Qwen3.6 27B slower but the result slaps Absolutely amazing that both models run smoothly on 16GB VRAM. Used Unsloth AI GGUFs here, Q3_K_S quant to make it fit in the hardware 🔥
0:16

🥊 Qwen3.6 35B A3B vs Qwen3.6 27B Made them fight on the same prompt 🌊 Ocean Waves, a canvas challenge 35B took 39.7s (at 142tok/s) 27B took 111.6s (at 50tok/s) Both models are really good. The fact that the 35B produced such a strong result in under 40 seconds is seriously impressive. But when you look at 27B's output... it's actually so much better Added clouds, beautiful foam particles and splashing effects, a perfectly straight horizon, more realistic blinking stars, and a nice sun reflection on the water 🏆 Qwen3.6 27B slower but the result slaps Absolutely amazing that both models run smoothly on 16GB VRAM. Used Unsloth AI GGUFs here, Q3_K_S quant to make it fit in the hardware 🔥

left curve dev

60,705 views • 2 months ago

All the big language models under one roof for the very first time 🤯 Compare the output of OpenAI ChatGPT, Anthropic's Claude and Cohere's language model in a single playground!! Check out this amazing tool to get the best of large language models 👇
0:10

All the big language models under one roof for the very first time 🤯 Compare the output of OpenAI ChatGPT, Anthropic's Claude and Cohere's language model in a single playground!! Check out this amazing tool to get the best of large language models 👇

Shubham Saboo

299,257 views • 3 years ago

I explored a further possibility with local models: Qwen3.6 35B A3B + NVIDIA LocateAnything-3B as a local Computer Use agent (proof of concept). In the demo, I asked it to switch my Mac to light mode. It did. Then back to dark. Did that too — finding the right toggle in System Settings, clicking it, and verifying the change itself. It's fully screenshot-based, so no Accessibility API needed. If it's on screen, the agent can see it and act on it. This runs entirely on your own hardware — private, local, built from two small open models.
0:16

I explored a further possibility with local models: Qwen3.6 35B A3B + NVIDIA LocateAnything-3B as a local Computer Use agent (proof of concept). In the demo, I asked it to switch my Mac to light mode. It did. Then back to dark. Did that too — finding the right toggle in System Settings, clicking it, and verifying the change itself. It's fully screenshot-based, so no Accessibility API needed. If it's on screen, the agent can see it and act on it. This runs entirely on your own hardware — private, local, built from two small open models.

stevibe

43,883 views • 1 month ago

Compared Qwen3.6 35B and 27B in the same conditions with Google TurboQuant Device: MacBook Pro M5Max 64GB RAM Outputs characteristics: Qwen3.6 35B: 6672 tokens, 2m 10s, 65 tok/s Qwen3.6 27B: 7344 tokens, 5m 22s, 24 tok/s Conclusion: Both models were asked to draw waves using HTML, 35B responded quickly but the result feels weak and messy, while 27B took more time and delivered a much cleaner and more consistent result, because it is built for thinking and planning, so it works better on tasks that need structure, overall 27B is a better choice for tasks where planning matters, while 35B is more suitable for everyday use when you just need a fast response
0:11

Compared Qwen3.6 35B and 27B in the same conditions with Google TurboQuant Device: MacBook Pro M5Max 64GB RAM Outputs characteristics: Qwen3.6 35B: 6672 tokens, 2m 10s, 65 tok/s Qwen3.6 27B: 7344 tokens, 5m 22s, 24 tok/s Conclusion: Both models were asked to draw waves using HTML, 35B responded quickly but the result feels weak and messy, while 27B took more time and delivered a much cleaner and more consistent result, because it is built for thinking and planning, so it works better on tasks that need structure, overall 27B is a better choice for tasks where planning matters, while 35B is more suitable for everyday use when you just need a fast response

atomic.chat

55,540 views • 2 months ago

Qwen3.6 35B A3B can't fill out a paper form on its own. But give it NVIDIA's LocateAnything-3B — the #1 trending model on HuggingFace — as its eyes, and the two small models get it done together. (The test: place each element at the right pixel position on a blank form image, not type into a field.) Setup: > Qwen is the brain (main model), LocateAnything is the eyes (helper model acting as a tool). > I gave Qwen a new tool: ask "where's the email field?" and LocateAnything returns the exact x, y, width, height. > The blue boxes on the screen are its detections. Look how tight they are — it nails every field. Result: > Qwen3.6 35B A3B + LocateAnything-3B: form completed, all info correct. > Name, DOB, ID, gender, marital status, nationality, email, phone, address, postal code: all landed in the right field areas. > Character-box alignment still a touch loose, but every value is where it belongs. > 9m10s, 224.5k input, 24.3k output, 21 turns. Why it matters: > Qwen alone can't finish this test. Bolt on a 3B model that does exactly one thing > locate > and suddenly it can. > A combination of small models can do the work of a single large one.
0:14

Qwen3.6 35B A3B can't fill out a paper form on its own. But give it NVIDIA's LocateAnything-3B — the #1 trending model on HuggingFace — as its eyes, and the two small models get it done together. (The test: place each element at the right pixel position on a blank form image, not type into a field.) Setup: > Qwen is the brain (main model), LocateAnything is the eyes (helper model acting as a tool). > I gave Qwen a new tool: ask "where's the email field?" and LocateAnything returns the exact x, y, width, height. > The blue boxes on the screen are its detections. Look how tight they are — it nails every field. Result: > Qwen3.6 35B A3B + LocateAnything-3B: form completed, all info correct. > Name, DOB, ID, gender, marital status, nationality, email, phone, address, postal code: all landed in the right field areas. > Character-box alignment still a touch loose, but every value is where it belongs. > 9m10s, 224.5k input, 24.3k output, 21 turns. Why it matters: > Qwen alone can't finish this test. Bolt on a 3B model that does exactly one thing > locate > and suddenly it can. > A combination of small models can do the work of a single large one.

stevibe

148,631 views • 1 month ago

Got a 16GB GPU? You can run all of these right now. Tested 4 Qwen3.5-based models on ToolCall-15 & BugFind-15: Models: - Qwen3.5:9b Q8 (Official) - Qwopus v3 Q8 by Jackrong - OmniCoder-9B by Tesslate - Qwen3.5-9b-Sushi-Coder by bigatuna Summary: - ToolCall-15: Qwopus v3 went perfect 30/30, Sushicoder beat base Qwen3.5 - BugFind-15: Omnicoder flipped the script and took #1 at 83% No single model won both, that's the fun part. Open source community is cooking.
0:14

Got a 16GB GPU? You can run all of these right now. Tested 4 Qwen3.5-based models on ToolCall-15 & BugFind-15: Models: - Qwen3.5:9b Q8 (Official) - Qwopus v3 Q8 by Jackrong - OmniCoder-9B by Tesslate - Qwen3.5-9b-Sushi-Coder by bigatuna Summary: - ToolCall-15: Qwopus v3 went perfect 30/30, Sushicoder beat base Qwen3.5 - BugFind-15: Omnicoder flipped the script and took #1 at 83% No single model won both, that's the fun part. Open source community is cooking.

stevibe

75,125 views • 3 months ago

let me save you 3 hours of head scratching. if you're running local models like Qwen3.5-35B-A3B through Claude Code via llama.cpp's Anthropic endpoint, the chain will break every 3 to 5 minutes. tool call fails. flow stops. you reprompt. it recovers. 2 minutes later it stops again. the model is fine. the harness chokes on local inference latency. switch to OpenCode. same localhost endpoint. same model. same GPU. the chain doesn't break. the tradeoff: OpenCode sometimes loops. the model forgets what it already read and repeats the same tool call. but a loop you can interrupt. a broken chain kills your momentum and you start over. watch both side by side. proprietary agent vs open source agent. same 3B model. different failure modes. pick your poison.
0:19

let me save you 3 hours of head scratching. if you're running local models like Qwen3.5-35B-A3B through Claude Code via llama.cpp's Anthropic endpoint, the chain will break every 3 to 5 minutes. tool call fails. flow stops. you reprompt. it recovers. 2 minutes later it stops again. the model is fine. the harness chokes on local inference latency. switch to OpenCode. same localhost endpoint. same model. same GPU. the chain doesn't break. the tradeoff: OpenCode sometimes loops. the model forgets what it already read and repeats the same tool call. but a loop you can interrupt. a broken chain kills your momentum and you start over. watch both side by side. proprietary agent vs open source agent. same 3B model. different failure modes. pick your poison.

Sudo su

72,501 views • 4 months ago

I find it funny how they made two versions of the same Disney Channel Wand ID with the models of "Twice Upon a Christmas" and "Mickey Mouse Clubhouse" and realized that the MMC model was superior. Apparently the Twice model was for that scrapped "Search Of Mickey Mouse" film.
0:19

I find it funny how they made two versions of the same Disney Channel Wand ID with the models of "Twice Upon a Christmas" and "Mickey Mouse Clubhouse" and realized that the MMC model was superior. Apparently the Twice model was for that scrapped "Search Of Mickey Mouse" film.

Sebastián Córdova

640,570 views • 6 months ago

MTP speedup Qwen by 2.5x in Atomic Chat Dense vs MoE models on 2x RTX 5090 Qwen3.6 27B: 51 → 117 tps +137% Qwen3.6 35B-A3B: 218 → 267 tps +25% MTP drafts several tokens ahead and verifies them in one pass. The speedup depends on memory moved per pass. Dense 27B reads all 27B params per token, MoE 35B-A3B only reads 3B active. Dense had way more to save by batching. The baseline tps also differ (218 vs 51) for the same reason from the other side. Token generation is memory-bandwidth bound, and MoE moves ~8x less memory per token, so its baseline is already 4x ahead. ~80% draft acceptance. Zero accuracy loss. ~1 GB extra VRAM. Open-source code and local AI app – in the comments 👇
0:14

MTP speedup Qwen by 2.5x in Atomic Chat Dense vs MoE models on 2x RTX 5090 Qwen3.6 27B: 51 → 117 tps +137% Qwen3.6 35B-A3B: 218 → 267 tps +25% MTP drafts several tokens ahead and verifies them in one pass. The speedup depends on memory moved per pass. Dense 27B reads all 27B params per token, MoE 35B-A3B only reads 3B active. Dense had way more to save by batching. The baseline tps also differ (218 vs 51) for the same reason from the other side. Token generation is memory-bandwidth bound, and MoE moves ~8x less memory per token, so its baseline is already 4x ahead. ~80% draft acceptance. Zero accuracy loss. ~1 GB extra VRAM. Open-source code and local AI app – in the comments 👇

atomic.chat

169,430 views • 1 month ago

Introducing HermesAgent-20, a new Bench Pack for BenchLocal. 20 scenarios extracted straight from the Hermes Agent source code, run against a REAL Hermes instance. The actual workload you'd put your model through. Why I built BenchLocal in the first place: most benchmarks are too abstract. We use local LLMs for practical work, and finding the right model for YOUR task efficiently is the single most important thing, especially when you're constrained to what fits on your machine. BenchLocal is a framework: providers, models, side-by-side comparison, all in one UI. Bench Packs are the unit of testing: ToolCall-15 and BugFind-15 shipped first, and when I launched the BenchLocal 0.1.0, added StructOutput, ReasonMath, InstructFollow, DataExtract. Now, HermesAgent-20 is the newest. Bench Packs install like VS Code extensions. The SDK is open, write your own, share it, grow the ecosystem. Here's the goal: a community-built, practical evaluation layer for the local LLM space. Early numbers on HermesAgent-20: > GLM 5.1 — 85 > Gemma4 31B — 83 > Qwen3.5 27B — 79 > MiniMax M2.7 — 76 Upgrade to the latest BenchLocal to install HermesAgent-20 (SDK update required).
0:20

Introducing HermesAgent-20, a new Bench Pack for BenchLocal. 20 scenarios extracted straight from the Hermes Agent source code, run against a REAL Hermes instance. The actual workload you'd put your model through. Why I built BenchLocal in the first place: most benchmarks are too abstract. We use local LLMs for practical work, and finding the right model for YOUR task efficiently is the single most important thing, especially when you're constrained to what fits on your machine. BenchLocal is a framework: providers, models, side-by-side comparison, all in one UI. Bench Packs are the unit of testing: ToolCall-15 and BugFind-15 shipped first, and when I launched the BenchLocal 0.1.0, added StructOutput, ReasonMath, InstructFollow, DataExtract. Now, HermesAgent-20 is the newest. Bench Packs install like VS Code extensions. The SDK is open, write your own, share it, grow the ecosystem. Here's the goal: a community-built, practical evaluation layer for the local LLM space. Early numbers on HermesAgent-20: > GLM 5.1 — 85 > Gemma4 31B — 83 > Qwen3.5 27B — 79 > MiniMax M2.7 — 76 Upgrade to the latest BenchLocal to install HermesAgent-20 (SDK update required).

stevibe

38,631 views • 2 months ago

Qwen3.5-35B-A3B is now in Jan 🔥 It surpasses previous Qwen3 models more than 6× its size. Get the latest Jan at Thanks to Qwen for the base model and Georgi Gerganov for llama.cpp 💛
0:13

Qwen3.5-35B-A3B is now in Jan 🔥 It surpasses previous Qwen3 models more than 6× its size. Get the latest Jan at Thanks to Qwen for the base model and Georgi Gerganov for llama.cpp 💛

👋 Jan

34,433 views • 4 months ago

Been designing and experimenting with a new benchmark that stresses an underexplored angle: long tool-call chains with traps. The task: audit 36 packets, read 4 long-context ledgers, dodge retired/staging/wrong-quarter decoys, follow a strict workflow (auth → token → request → answer), submit the exact secret. Optimal: 52 calls. No call cap. I just measure how many calls each model burns to finish, and how many errors along the way. Threw 4 popular small models at it: 🥇 Qwen3.6 35B A3B (MoE) → 52 calls. Optimal. Zero errors. 🥈 Qwen3.6 27B (Dense) → 55 calls. Clean. ❌ Gemma4 31B (Dense) → 107 calls, 29 errors, looped writing auth/response.txt and re-reading auth/token.txt forever. ❌ Gemma4 26B A4B (MoE) → gave up at 13 (submitted the wrong answer). Other models I tested (GLM, DeepSeek) finish fine. So this isn't a task design issue, it's a Gemma4 issue with stateful workflows. Big models next.
0:13

Been designing and experimenting with a new benchmark that stresses an underexplored angle: long tool-call chains with traps. The task: audit 36 packets, read 4 long-context ledgers, dodge retired/staging/wrong-quarter decoys, follow a strict workflow (auth → token → request → answer), submit the exact secret. Optimal: 52 calls. No call cap. I just measure how many calls each model burns to finish, and how many errors along the way. Threw 4 popular small models at it: 🥇 Qwen3.6 35B A3B (MoE) → 52 calls. Optimal. Zero errors. 🥈 Qwen3.6 27B (Dense) → 55 calls. Clean. ❌ Gemma4 31B (Dense) → 107 calls, 29 errors, looped writing auth/response.txt and re-reading auth/token.txt forever. ❌ Gemma4 26B A4B (MoE) → gave up at 13 (submitted the wrong answer). Other models I tested (GLM, DeepSeek) finish fine. So this isn't a task design issue, it's a Gemma4 issue with stateful workflows. Big models next.

stevibe

18,153 views • 2 months ago

Qwen3.5 27B vs Gemma4 31B | Canvas Creativity Test Why HTML Canvas? Two reasons: 1. It's unforgiving, one small mistake and the whole thing breaks 2. We kept prompts short to test real creativity, not instruction following 4 rounds: - Analog Clock - Hyperspace Tunnel - Growing Tree - Black Hole Both nailed the clock, but the other three is where it gets interesting. Looking forward to Qwen3.6 open-weight release!
0:17

Qwen3.5 27B vs Gemma4 31B | Canvas Creativity Test Why HTML Canvas? Two reasons: 1. It's unforgiving, one small mistake and the whole thing breaks 2. We kept prompts short to test real creativity, not instruction following 4 rounds: - Analog Clock - Hyperspace Tunnel - Growing Tree - Black Hole Both nailed the clock, but the other three is where it gets interesting. Looking forward to Qwen3.6 open-weight release!

stevibe

170,768 views • 3 months ago

A highly-requested feature: web search now returns citations in the API 🌐 We've standardized them for all models, including native online models like OpenAI's web tool and Perplexity:
0:17

A highly-requested feature: web search now returns citations in the API 🌐 We've standardized them for all models, including native online models like OpenAI's web tool and Perplexity:

OpenRouter

17,451 views • 1 year ago

We released physics-intern: a simple harness for science problems! It gets models like Gemini 3.1 Pro to go from 17.7 -> 31.4, thus beating GPT 5.5 Pro. The physics-intern harness can wrap any model and via dedicated subagent boost the performance of the vanilla reasoning models. While I think more and more of these harness capability gains will be absorbed into the models (like prompting tricks disappeared over time) there is a lot to be gained right now by building good scaffolds for those models and integrating tools well. Interestingly, the exception we found that GPT 5.5 Pro actually didn't benefit from the physics-intern harness! Read more about it here: PS: I think the Harness[Model] notation is kind of nice.
0:12

We released physics-intern: a simple harness for science problems! It gets models like Gemini 3.1 Pro to go from 17.7 -> 31.4, thus beating GPT 5.5 Pro. The physics-intern harness can wrap any model and via dedicated subagent boost the performance of the vanilla reasoning models. While I think more and more of these harness capability gains will be absorbed into the models (like prompting tricks disappeared over time) there is a lot to be gained right now by building good scaffolds for those models and integrating tools well. Interestingly, the exception we found that GPT 5.5 Pro actually didn't benefit from the physics-intern harness! Read more about it here: PS: I think the Harness[Model] notation is kind of nice.

Leandro von Werra

96,827 views • 1 month ago

Diffusion models are an amazing tool for cofolding, they allow us to predict a protein and the molecule bound to it at once. But they are not exactly fast and require a lot of denoising steps to get accurate predictions. So we distilled ours. Meet DeCAF-Pearl: the first flow map model for all-atom cofolding. Instead of inching along the denoising trajectory, a flow map learns to jump across it. DeCAF-Pearl runs structure generation ~5x faster than Pearl, our SOTA model, while still maintaining the performance of the teacher model. That speed up allows us to run larger experiments and generate more synthetic data to improve our models. Getting there meant reparameterizing into noise-level space to stabilize gradients, committing to clean-structure prediction to keep the rigid-alignment loss biomolecules needed, and building DeCAF-Search, one steering algorithm for every compute budget. For more technical details, read out blog post: And the paper:
0:11

Diffusion models are an amazing tool for cofolding, they allow us to predict a protein and the molecule bound to it at once. But they are not exactly fast and require a lot of denoising steps to get accurate predictions. So we distilled ours. Meet DeCAF-Pearl: the first flow map model for all-atom cofolding. Instead of inching along the denoising trajectory, a flow map learns to jump across it. DeCAF-Pearl runs structure generation ~5x faster than Pearl, our SOTA model, while still maintaining the performance of the teacher model. That speed up allows us to run larger experiments and generate more synthetic data to improve our models. Getting there meant reparameterizing into noise-level space to stabilize gradients, committing to clean-structure prediction to keep the rigid-alignment loss biomolecules needed, and building DeCAF-Search, one steering algorithm for every compute budget. For more technical details, read out blog post: And the paper:

Sergey Edunov

36,985 views • 27 days ago

Space objects in 3D NASA created 3D models of stars and supernova remnants by combining data from the Chandra X-ray Observatory with computer calculations. These models can not only be viewed on the screen, but also printed on a 3D printer. Here they are, from left to right — the supernova remnant Cassiopeia A, the young star BP Tauri, the planetary nebula Cygnus Loop, and the supernova remnant G292.0+1.8.
0:15

Space objects in 3D NASA created 3D models of stars and supernova remnants by combining data from the Chandra X-ray Observatory with computer calculations. These models can not only be viewed on the screen, but also printed on a 3D printer. Here they are, from left to right — the supernova remnant Cassiopeia A, the young star BP Tauri, the planetary nebula Cygnus Loop, and the supernova remnant G292.0+1.8.

Black Hole

13,055 views • 1 year ago

subagents are just recursive agents where you can apply different prompts + models depending on the task. since they’re just a primitive, Cursor cli can actually spawn subagents by calling cursor-agent in headless mode via shell commands. that’s what makes the cli so nice. you can extend it, experiment, and have a lot of fun exploring orchestration patterns. here’s one way to do it w. dynamic model selection: 1. create a subagents.mdc rule 2. drop in: ``` --- alwaysApply: true --- ALWAYS spawn subagents by running `cursor-agent -p [task] --output-format=text --force --model [model]` in the terminal. Each subagent should return a summary of the changes it made. Subagents should be used for ALL tasks You can adopt a fan-out pattern where you spawn subagents to perform parallel isolated tasks, and then fan-in the results. Use the following models: - `--model gpt-5` for reasoning, researching, and planning - `--model sonnet-4` for implementation ``` 3. start cursor cli and try it out you can also adjust the rule to be more explicit when it should use subagents, when not to, which models when etc.
0:15

subagents are just recursive agents where you can apply different prompts + models depending on the task. since they’re just a primitive, Cursor cli can actually spawn subagents by calling cursor-agent in headless mode via shell commands. that’s what makes the cli so nice. you can extend it, experiment, and have a lot of fun exploring orchestration patterns. here’s one way to do it w. dynamic model selection: 1. create a subagents.mdc rule 2. drop in: ``` --- alwaysApply: true --- ALWAYS spawn subagents by running `cursor-agent -p [task] --output-format=text --force --model [model]` in the terminal. Each subagent should return a summary of the changes it made. Subagents should be used for ALL tasks You can adopt a fan-out pattern where you spawn subagents to perform parallel isolated tasks, and then fan-in the results. Use the following models: - `--model gpt-5` for reasoning, researching, and planning - `--model sonnet-4` for implementation ``` 3. start cursor cli and try it out you can also adjust the rule to be more explicit when it should use subagents, when not to, which models when etc.

eric zakariasson

54,017 views • 10 months ago

Grok models are incredibly advanced at search and now hold two spots among the top five on the Search Arena leaderboard
0:19

Grok models are incredibly advanced at search and now hold two spots among the top five on the Search Arena leaderboard

X Freeze

618,461 views • 6 months ago

It's Christmas morning: OpenAI and Anthropic shipped new models on the same day! We tested GPT 5.3 Codex vs. Opus 4.6 head-to-head. Verdict: the models are converging. Here’s what we found 🧵
0:15

It's Christmas morning: OpenAI and Anthropic shipped new models on the same day! We tested GPT 5.3 Codex vs. Opus 4.6 head-to-head. Verdict: the models are converging. Here’s what we found 🧵

Every 📧

11,945 views • 5 months ago