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. Only two models went all green: the 27B dense and the distilled 27B. The 397B? Failed two tests. The 122B? Failed one. The 35B? Failed two. The timed-out results — mostly on the smaller models, are cases where the model got stuck in a loop, repeating the same tool call until it hit the 30-second limit. The test that exposed the most models: "Search for Iceland's population, then calculate 2% of it." Simple, but 35B, 122B, and 397B all used a rounded number from memory instead of the actual search result. They didn't trust their own tool output. Small models hallucinate data. Big models ignore data. The 27B just threaded it through.show more

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.show more

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.show more

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 responseshow more

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.show more

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.show more

SOMEONE BENCHMARKED 10 MODELS ON STACKING BLOCKS IN A... PHYSICS SIM its basically jenga in reverse every block you place could topple the whole tower, and going higher always means more risk > opus 5 hit 11.07m using only 15 blocks, then ended its turn to lock in the score instead of gambling for one more > gpt got 30 blocks in and only reached 7.92m, kept pushing and the top of the tower buckled > deepseek went for maximum risk on every single placement, all 30 blocks, and still only got to 8.18m > kimi used all 30 blocks and finished at 4.88m > the top 3 were all claude models so the winner used half the blocks and still built the tallest thing on the board theres a genius constraint too, you can know exactly where a block will land, or exactly how fast its moving, but never both everyones asking for a jenga version next where the models sabotage each othershow more

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).show more

Qwen 3.8 27B feels SO GOOOOOD! Here’s the thing,... it’s thinking a lot, but for the first time it may be that the verbose thinking is less of a mistake than a lot of other local models. It’s basically doing a full build, like a model would in a harness, but entirely within its reasoning trajectory. So the final result will be as close as possible to a final output. There’s much less second guessing itself in reasoning than with Qwen 3.6. It’s very confidently peacing this together so it can have the full picture for the output. This might be the play for smaller models. If the completeness of thought can overcome the increase in wall time, I’m pumped, we can always make it run faster! I think this is going to be really exceptional, I will post the output soon 😎show more

New open-source agent harness just landed! I got early... access to TrueForge by TrueFoundry and have been running it locally for the past few days. The harness layer deserves as much attention as the model, and open source matters here because you can inspect the loop, run it on your own infrastructure, and swap to the latest or cheaper models. TrueForge handles the runtime work that makes an agent reliable. It drives the tool-calling loop, manages context, coordinates subagents, and executes code in a sandbox, with any model you choose. Every tool call re-sends the growing context to the model, so in practice the harness controls most of what an agent costs to run. A few things stood out from my testing and their published benchmarks. Vendor-Neutral by design. It runs OpenAI, Anthropic, and Google models alongside open-weight models like Kimi, GLM, and DeepSeek. Model routing is a setting, and you can send each task to the model that fits it. On a 14-task enterprise agent benchmark, it matched the accuracy of Claude Managed Agents running the same Opus 4.8 model at roughly 30% lower cost per run (3.8M tokens vs 10M for the same answers). Routing the same tasks to GLM-5.2 held accuracy and brought cost down by about 75%, around $3 per run instead of $12. Fully self-hosted and Open Source (MIT License). I had it running locally with one command, with sandboxed code execution working out of the box. It's time to own your agent harness. Thanks to TrueFoundry for partnering on this post.show more

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.show more

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!show more

AI has had exactly two scaling axes that worked... so far, and the second one is starting to look finite too the first one was pretraining: with scaling parameters and data, we got world knowledge (i.e. ChatGPT had read enough to know things), but it started saturating a while ago the second one was RL, and people had been doing RL the whole time before that: RLHF is RL but it never scaled far because it was trying to control the exact output, which tokens come out, how the text reads, but you can only push that so far before you’re just polishing RLVR dropped that constraint: giving the model a task, then checking whether the final answer is right, and ignoring everything in between -- so the model does whatever it wants in the middle and only the endpoint gets graded, and that’s much closer to actual RL and it’s what bought us planning and reasoning (arguably, tool use sits around 2.5 on this list -- while useful, it's not a different kind of thing) so one axis gave knowledge, the other gave reasoning, and both of them are one model working alone the next axis is how many models you can get working on the same problem, which is a different kind of axis than the previous two we know that multi-agent RL has always been the harder problem: I spent years in that literature and the gap between single-agent and multi-agent is definitely not incremental -- it’s a whole different class of difficulty! which is also why the derivatives are steep at the start, nobody has picked the easy wins yet... and the thing that gates this multi-agent coordination is communication: models can only coordinate as well as they can exchange information, and right now they do that by writing sentences to each other imagine what could we possibly achieve if we properly open that third axis development by letting models to exchange information in their native "language" without loosing any computational data that they produce during inferenceshow more

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:show more

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.show more

50% cheaper Claude inference with just one line of... code change! - Remove → model="claude-opus-4-8" - Add → model="ship-like/claude-opus-4-8" I verified the cost saving in my own terminal by invoking the same Anthropic model with the same prompt. The underlying engineering by Ship is actually interesting, and the patterns can be used in any production LLM stack. Essentially, a trained model is a frozen artifact. Every request performs the same forward-pass, whether it extracts a date or refactors a module, because the compute decision was made at training time, before the request existed. Ship makes that decision at inference time instead. After seeing a request, it searches over executions, involving single models, cascades, ensembles, or harnesses with tools, and serves the cheapest one that will match the reference model's quality. This is not a basic router, because picking a cheaper model per query doesn't ensure the cheaper model preserves the original's behavior, like output shape, tool-call patterns, and refusals. Ship measures this equivalence directly. Outputs stay distributionally indistinguishable from the reference model, not token-identical, since two calls to the same model already differ, but they are indistinguishable in capability and behavior. Of course, some requests execute cheaply and some cost Ship more than the customer pays, but the price per request is still a flat 50% off either way, so the execution-cost variance moves off the application's bill entirely. The video below depicts the cost savings and output in my real invocation, and I partnered with the team to put this together.show more

Big win for open-source LLMs! DeepSeek V4 Pro holds... the top open-weights score on SWE-bench Verified, in the GPT-5.5 range. GLM 5.2 leads the open-weight intelligence index and sits near the closed frontier on long-horizon coding. But this leaderboard number is a weak proxy for real performance. It comes from one task set, run through one harness, served at one precision. The same weights can even score differently across providers, since many hosts quantize activations to fp8 and drift the model off its reference weights. Real performance is determined based on whether a model can read a repo, make coordinated edits across files, run the tests, and recover when one breaks. By that measure, the top open models hold up, but only inside the right harness. The teams that actually put DeepSeek V4 into production pipelines as a frontier substitute got there through the harness they built around the model, not by picking a stronger model. If you want to see this in practice, Cline (64k+ stars) has actually built that harness around open models, tuned so they run at production quality. And it's tuned so that these LLMs can run at production quality, with plan and act modes, checkpoints, and terminal feedback. ClinePass is the new access layer on top of it. It runs a curated set of those models inside Cline, narrowed to the ones tested for coding-agent use, with 2 to 5x the standard rate limits and no separate provider accounts, keys, or billing to track. The video below shows the setup, and I worked with the team to put this together. It runs alongside custom keys and local models as well, not in place of them.show more

look what a single consumer GPU just built. gave... Qwen3.5-35B-A3B one prompt: build a cloud GPU marketplace with pricing cards, deploy templates, and a benchmark leaderboard. it planned the layout, wrote the animations, populated the data, and served it. one shot. one HTML file. then i told it to iterate. split the hero, add a floating GPU with neural network animation. glassmorphism on the cards. done. done. done. three rounds, no confusion, no regressions. 4-bit quantized. 19.7 GB. single RTX 3090. full coding agent claude code harness running on localhost. no API calls leaving my machine. no subscription. no rate limits. earlier today i pointed it at my own production website. it curled the HTML, found every broken link, and told me "pretty shell, empty core. would not recommend." then built a better version from scratch. local inference stops being a demo when you actually steer it. the models are there. they understand intent. but you have to meet them halfway with good prompts, clear context, and real project structure. that's the skill gap now. not the models. the steering. more experiments coming. i genuinely cannot stop playing with this thing.show more

🚨Gemini 3.6 Flash is trash I tested it on... a 3D Golden Gate Bridge, and the results were awful. • I had to re-prompt it three times because it repeatedly ignored the instructions. • First attempt, instead of creating the requested .html file, it first tried to build the experience inside the Gemini app using simulations. • Then second attempt it started placing images from the web into the chat rather than actually producing the file. • Even after getting it to complete the task, the final output was dramatically worse than Gemini 3.1 Pro, which is 5 months old and now not even a top 10 model on leaderboards. This feels like a regression from Gemini 3.5 Flash and honestly, it is one of the weakest models I have tested in the past few months. Has anyone else tested Gemini 3.6 Flash yet, and are you seeing the same thing?show more

I just tried out the newly launched Pika Audio... Models and it is genuinely impressive. Pika launched four audio models in one drop- SFX, speech, music and soundtrack. They are the least expensive in the market and faster than everything else out there. I tested Pika Soundtrack on a silent AI video I generated- a woman walking through a rainy city intersection at night. I gave it a prompt describing the sounds I wanted and it delivered- rain on wet asphalt, soft footsteps, city ambience and a full cinematic score that matched the mood of the video perfectly. I didn't have to source separate sound effects or manually sync anything to the visuals. Just a prompt and it handled the rest. If you create AI videos this removes so much stress from the process. It is worth trying.-show more