Open-Source Multi-Model Orchestration

Reasoning in Steps, Not Guesses.

TachiBot runs your prompt through structured reasoning tools — council, jury, planner, step-by-step — so models debate, verify, and reason deeper before they answer. One answer you can trust.

Gateway:One OpenRouter key

BYOB:Your own provider keys

Perplexity: Always needs own key

Get Started →Read Docs

"Have Grok check Twitter for that error message""Ask Perplexity what changed in React 19 this week""Get Gemini to brainstorm, then have Kimi K2.6 and GPT-5.5 both analyze it"

Providers

Tools

Prompt Techniques

Stop Trusting. Start Verifying.

Same question. Different approach. Better answer.

"What breaking changes are in React 19?"

One Model

Generic advice, might miss recent updates

No sources or documentation links

Could confuse React 18 vs 19 features

No way to verify accuracy

Unverified, possibly outdated

TachiBot

1. Run in Parallel

OpenAIGooglePerplexityOpenRouter

2. Cross-Verify

Models challenge each other, 3-200 rounds

3. Fact-Check

PerplexityGrok

Live sources with recency filters

Verified, with official sources

Get Started →

Building Blocks for Thinking

Mix and Match Your Reasoning Pipeline

Mix and match tools, prompt techniques, and workflows to build your reasoning pipeline.

New

Multi-Model Planner

Create bulletproof implementation plans. A council of 6 AI models — Grok searches for ground truth, Qwen analyzes code, Kimi K2.6 reasons step-by-step and decomposes into dependency-ordered subtasks, GPT-5.5 critiques gaps, Qwen 235B drafts the synthesis, and Gemini judges the final plan with quality scores and verification checkpoints.

planner_maker + planner_runner

New

Multi-Model Jury

Run your question through a lab-diverse panel of AI jurors in parallel. Grok reasons from first principles, DeepSeek brings frontier math & logic, Kimi decomposes step-by-step, GPT evaluates tradeoffs — with 12 jurors to choose from, including a free offline local juror. Gemini Judge synthesizes a unified verdict using science-backed LLM-as-a-Judge methodology.

jury + gemini_judge

12 AI Providers

GPT-5.5, Gemini, Grok, Perplexity, Kimi K2.6, Qwen, MiniMax, DeepSeek V4, GLM-5.1, StepFun, ERNIE — plus free local models via Ollama or LM Studio. Gateway mode with one OpenRouter key, or BYOB with your own provider keys.

>70%Qwen Coder-Next·SWE98%Qwen 235B·HMMT80.2%Kimi K2.6·SWE93.6%GPT-5.5·GPQA78.8%Gemini 3.5 Flash·SWE56.2%MiniMax M2.7·SWE-Pro

gateway + BYOB + local

New

31 Prompt Techniques

Research-backed patterns including reflexion, SCoT, ReAct, pre_mortem, and 27 more. Embedded in tool prompts for automatic application. Preview before executing.

preview → execute

YAML Workflows

Chain unlimited steps. Variable interpolation, step dependencies, auto-distillation, comparison tables, and optional AI judge. PingPong debates, research pipelines, code reviews.

workflow + planner

Multi-model deliberation

The Council & Jury

One question fans out to a panel of independent models in parallel. Their perspectives are aggregated, then a Gemini judge synthesizes a single verdict — the “replace judges with juries” pattern, made real.

FLOW · jury → gemini_judge

—·Press replay to trace the deliberation

QUERY

grok

first-principles

deepseek

frontier math

kimi

step-by-step

openai

tradeoffs

AGGREGATE

perspectives

gemini

presiding judge

VERDICT

active data path jurors (parallel) presiding judge

Where each model is strongest

researched via grok_search + perplexity_ask · Jun 2026

Grok 4.3juror

Real-time info + cost-efficient agentic tasks; lowest hallucination rate.

DeepSeek V4 Projuror

Open-weight frontier reasoning — top AIME & CodeElo math/CP scores.

Kimi K2.6juror

Scalable agent swarms & long-horizon tool use — SWE-bench Pro leader.

GPT-5.5juror

Agentic workflows & rock-solid tool-use reliability.

Gemini 3.1 Projudge

Long-context multimodal reasoning & science — the presiding judge.

GLM-5.1

SWE-Bench Pro leader; agentic tool-use & planning juror.

StepFun 3.7 Flash

Near-frontier AIME/SWE-Verified reasoning at flash-tier cost.

ERNIE 4.5 VL

Broad world knowledge & human-preference/arena strength.

Gemini 3.5 Flash

High-speed frontier agentic coding & grounded search.

Qwen3-Max-Thinking

Heavy mathematical & formal reasoning in thinking mode (HMMT 98%).

MiniMax M2.7

Extreme long context + self-evolving agents, cost-efficient.

Local models

Free offline juror via Ollama / LM Studio — zero token cost, fully private. Runs any local model, incl. Nous Hermes builds. (The Hermes agent itself is model-agnostic — 300+ backends from GPT and Claude to self-hosted.)

Judge

gemini_judge

One model · one pass

A single evaluator. Give it a set of perspectives you already have and it scores, ranks, resolves conflicts, or merges them into one answer. It is the synthesis primitive — the last step the jury and council both lean on.

Jury

jury

Many models · one round

Many jurors answer in parallel, then a judge synthesizes. One tool call, fast, and the default entry point for any “which is right / which is best?” question. = N jurors + 1 judge.

Council

/council

Many models · many rounds

The heavyweight. Research → adversarial debate → sequential reasoning → final synthesis, iterated over multiple rounds. For architecture calls and deep trade-offs where one round isn't enough.

The arsenal

Every tool, filterable

Filter the full toolset by capability or provider — or search by name. Each tool follows the strict design rule: one required string content param, all enums optional with defaults.

Capability

Provider

Showing 57 tools

grok_reasonGrok (xAI)

Deep first-principles reasoning with optional heavy mode.

Reasoninggrok-4.3

openai_reasonOpenAI

Mathematical & high-effort reasoning with GPT-5.5 thinking.

Reasoninggpt-5.5

qwen_reasonQwen

Heavy mathematical reasoning (235B thinking).

Reasoningqwen3-coder-next

qwq_reasonQwen

Multi-perspective deliberation: optimist, pessimist, expert, contrarian.

Reasoningqwen3-coder-next

kimi_thinkingKimi

Multimodal reasoning with a 100-subagent swarm.

Reasoningkimi-k2.6

perplexity_reasonPerplexity

Reasoning grounded in live web search.

Reasoningsonar

deepseek_reasonDeepSeek

Open-weight frontier reasoning — top AIME/GPQA math & logic chains.

Reasoningdeepseek-v4-pro

glm_reasonGLM (Z.ai)

Agentic reasoning & tool-use planning (SWE-Bench Pro leader).

Reasoningglm-5.1

stepfun_reasonStepFun

Efficient deep reasoning — high AIME/SWE-Verified at flash-tier cost.

Reasoningstep-3.7-flash

ernie_reasonERNIE (Baidu)

Broad-knowledge reasoning with human-preference/arena strength.

Reasoningernie-4.5-vl

local_queryLocal

Query a local model via Ollama/LM Studio/llama.cpp/vLLM — free, offline, private.

Reasoningollama · lm studio

nextThoughtTachiBot

Sequential thinking chain with optional per-step model execution.

Reasoningorchestration

thinkTachiBot

Lightweight internal reasoning log.

Reasoningorchestration

qwen_coderQwen

Agentic code generation. SWE-bench >70%.

Codingqwen3-coder-next

qwen_algoQwen

Algorithm analysis: complexity & optimization tiers.

Codingqwen3-coder-next

deepseek_algoDeepSeek

Strongest algorithmic review: correctness, Big-O, edge cases, data structures.

Codingdeepseek-v4-pro

qwen_competitiveQwen

Competitive-programming problem solving.

Codingqwen3-coder-next

kimi_codeKimi

SWE-focused code generation & bug-fixing leader.

Codingkimi-k2.6

minimax_codeMiniMax

SWE-Pro code operations, cost-efficient (56% SWE-Pro).

Codingminimax-m2.7

grok_codeGrok (xAI)

Code analysis & optimization.

Codinggrok-4.3

grok_debugGrok (xAI)

Specialized debugging assistance.

Codinggrok-4.3

openai_code_reviewOpenAI

Comprehensive code review with focus areas.

Codinggpt-5.5

perplexity_askPerplexity

Web search with current information (Sonar).

Researchsonar

perplexity_researchPerplexity

Deep research — synthesizes hundreds of sources into a report.

Researchsonar

grok_searchGrok (xAI)

Live web search with Grok.

Researchgrok-4.3

openai_searchOpenAI

Real-time web search via GPT-5.5.

Researchgpt-5.5

gemini_searchGemini

Web search with Google Search grounding.

Researchgemini-3.5-flash

gemini_analyze_codeGemini

Code quality, security & performance analysis.

Analysisgemini-3.1-pro · 3.5-flash

gemini_analyze_textGemini

Rhetorical analysis: bias, fallacies, persuasion.

Analysisgemini-3.1-pro · 3.5-flash

kimi_long_contextKimi

Long-context analysis across a 256K window.

Analysiskimi-k2.6

openai_explainOpenAI

Clear explanations for complex topics, any level.

Analysisgpt-5.5

grok_brainstormGrok (xAI)

Contrarian first-principles idea generation.

Brainstormgrok-4.3

openai_brainstormOpenAI

Find the 3rd, 4th & 5th alternative approaches.

Brainstormgpt-5.5

gemini_brainstormGemini

Convergent synthesis — clusters ideas into themes.

Brainstormgemini-3.1-pro · 3.5-flash

juryTachiBot

Multi-model jury panel → Gemini judge synthesis.

Councilorchestration

gemini_judgeGemini

Multi-perspective evaluation & verdict synthesis.

Councilgemini-3.1-pro · 3.5-flash

planner_makerTachiBot

Council-based plan creation via coordinator pattern.

Planningorchestration

planner_runnerTachiBot

Execute plans against goal-oriented verification gates.

Planningorchestration

list_plansTachiBot

List recently created plans.

Planningorchestration

kimi_decomposeKimi

Structured task decomposition via agent swarm.

Planningkimi-k2.6

minimax_agentMiniMax

Multi-step task decomposition & execution.

Planningminimax-m2.7

grok_architectGrok (xAI)

System architecture & design via 4–16 agent swarm.

Planninggrok-4.3

workflowTachiBot

Execute YAML-based multi-step workflows.

Workfloworchestration

workflow_startTachiBot

Begin a workflow session with variables.

Workfloworchestration

continue_workflowTachiBot

Resume an interrupted workflow session.

Workfloworchestration

list_workflowsTachiBot

Discover available workflows.

Workfloworchestration

create_workflowTachiBot

Create a new YAML workflow definition.

Workfloworchestration

workflow_statusTachiBot

Check progress of a running workflow session.

Workfloworchestration

validate_workflowTachiBot

Validate workflow YAML: syntax, tool registry, dependency graph.

Workfloworchestration

validate_workflow_fileTachiBot

Validate a workflow file on disk before running.

Workfloworchestration

visualize_workflowTachiBot

Render a workflow's step graph for inspection.

Workfloworchestration

list_prompt_techniquesTachiBot

Discover all 31 prompt-engineering techniques.

Prompt Eng.orchestration

preview_prompt_techniqueTachiBot

Preview an enhanced prompt before executing.

Prompt Eng.orchestration

execute_prompt_techniqueTachiBot

Run a query with a chosen technique applied.

Prompt Eng.orchestration

tachiTachiBot

Smart auto-router: research, solve, verify, create, judge.

Routingorchestration

focusTachiBot

Mode-based multi-model reasoning (deep, debate, analyze).

Routingorchestration

usage_statsTachiBot

View or reset tool usage & cost statistics.

Routingorchestration

The roster

Twelve providers, one interface

Every model is wired through a single MCP surface with automatic fallbacks. Tools register only when their API key is present — including free offline local models.

OpenAI

5 tools

gpt-5.5 · 1.1M ctx

Agentic reasoning, code review & explanations. The reliable workhorse for synthesis.

Gemini

5 tools

gemini-3.1-pro · 3.5-flash · 1M ctx

Multi-perspective analysis, brainstorming & the presiding jury judge on 3.1 Pro; search runs the agentic 3.5 Flash.

Grok (xAI)

6 tools

grok-4.3 · 1M ctx

First-principles reasoning, architecture swarms & contrarian brainstorming. Lowest hallucination rate, 1M context.

Kimi

4 tools

kimi-k2.6 · 256K ctx

Task decomposition, SWE coding & long-context analysis via a 1T-param MoE agent swarm.

Qwen

5 tools

qwen3-coder-next · 262K ctx

Algorithm analysis, agentic code generation & heavy mathematical reasoning. SWE >70%.

MiniMax

2 tools

minimax-m2.7 · MoE 230B

Cost-efficient SWE-Pro coding & multi-step agentic execution. Multi-SWE #1.

Perplexity

3 tools

sonar · live web

Live web search & research grounded in current sources. Cost-optimized on Sonar.

DeepSeek

2 tools

deepseek-v4-pro · open wts

Open-weight frontier reasoning — top AIME/CodeElo math & competitive programming. Default juror & strongest algorithmic reviewer.

GLM (Z.ai)

1 tools

glm-5.1 · agentic

SWE-Bench Pro leader. Agentic reasoning & tool-use planning from Zhipu.

StepFun

1 tools

step-3.7-flash · 196B

Efficient deep reasoning — near-frontier AIME/SWE-Verified scores at flash-tier cost.

ERNIE (Baidu)

1 tools

ernie-4.5-vl · 424B MoE

Broad world knowledge & human-preference strength. Multimodal 424B/47B-active MoE.

Local

1 tools

ollama · lm studio · offline

Any OpenAI-compatible local server — Ollama, LM Studio, llama.cpp, vLLM. Zero-cost, private, offline jurors.

See how it works — Routing, Planner & Internals →

Get Started in Minutes

Add to Claude Desktop or any MCP client

1Install via npm

npm install -g tachibot-mcp

2Add to Claude Desktop config

{
  "mcpServers": {
    "tachibot": {
      "command": "tachibot-mcp",
      "args": ["--config", "~/.tachibot/config.json"]
    }
  }
}

~/.config/Claude/claude_desktop_config.json

3Start using

tachibot workflow run general-council "Your query here"

Documentation GitHub

Built in public. Backed by stars.

TachiBot is open source and actively maintained. If it helps your workflow, a star helps us keep going.

Star on GitHub Report a bug or request a feature