diff --git a/README.md b/README.md
index d7c80edc..9a53da8e 100644
--- a/README.md
+++ b/README.md
@@ -146,7 +146,6 @@ import { run } from "@perstack/runtime"
 
 const checkpoint = await run({
   setting: {
-    model: "claude-sonnet-4-5",
     providerConfig: { providerName: "anthropic", apiKey: env.ANTHROPIC_API_KEY },
     expertKey: "my-expert",
     input: { text: query },
diff --git a/docs/getting-started/walkthrough.mdx b/docs/getting-started/walkthrough.mdx
index e11f558c..eaa1ed0e 100644
--- a/docs/getting-started/walkthrough.mdx
+++ b/docs/getting-started/walkthrough.mdx
@@ -276,7 +276,6 @@ import { run } from "@perstack/runtime"
 
 const checkpoint = await run({
   setting: {
-    model: "claude-sonnet-4-5-20250929",
     providerConfig: { providerName: "anthropic" },
     expertKey: "fitness-assistant",
     input: { text: "Start today's session" },
@@ -291,7 +290,6 @@ import { run } from "@perstack/runtime"
 
 const checkpoint = await run({
   setting: {
-    model: "claude-sonnet-4-5-20250929",
     providerConfig: { providerName: "anthropic" },
     expertKey: "fitness-assistant",
     input: { text: "Start today's session" },
diff --git a/docs/operating-experts/deployment.md b/docs/operating-experts/deployment.md
index 6170e7b8..2c19085e 100644
--- a/docs/operating-experts/deployment.md
+++ b/docs/operating-experts/deployment.md
@@ -92,7 +92,6 @@ export default {
     await run(
       {
         setting: {
-          model: "claude-sonnet-4-5",
           providerConfig: { providerName: "anthropic", apiKey: env.ANTHROPIC_API_KEY },
           expertKey,
           input: { text: query },
diff --git a/docs/references/cli.md b/docs/references/cli.md
index 8cd2ccbd..09b72e6f 100644
--- a/docs/references/cli.md
+++ b/docs/references/cli.md
@@ -74,7 +74,7 @@ Both `start` and `run` accept the following options:
 | Option                        | Default         | Description                                                                            |
 | ----------------------------- | --------------- | -------------------------------------------------------------------------------------- |
 | `--provider <provider>`       | `anthropic`     | LLM provider                                                                           |
-| `--model <model>`             | `claude-sonnet-4-5` | Model name                                                                         |
+| `--model <model>`             | `auto`              | Model name (auto-resolved from expert tier or provider's middle tier)               |
 | `--reasoning-budget <budget>` | -               | Reasoning budget (`minimal`, `low`, `medium`, `high`, or token count)                  |
 
 Providers: `anthropic`, `google`, `openai`, `deepseek`, `ollama`, `azure-openai`, `amazon-bedrock`, `google-vertex`
diff --git a/docs/references/providers-and-models.md b/docs/references/providers-and-models.md
index 3f642e80..4f256789 100644
--- a/docs/references/providers-and-models.md
+++ b/docs/references/providers-and-models.md
@@ -8,12 +8,12 @@ Perstack supports multiple LLM providers. Configure via CLI options, environment
 
 ## Default Model
 
-Perstack uses `claude-sonnet-4-5` as the default model, selected based on:
+When no model is specified, Perstack resolves it automatically using a tier-based system:
 
-- **Standard tier pricing** — not a flagship model, economically sustainable for extended runs
-- **High agentic performance** — demonstrated tool use capability in benchmarks like [𝜏²-Bench](https://artificialanalysis.ai/evaluations/tau2-bench)
+1. **Expert's `defaultModelTier`** — if the expert definition sets a `defaultModelTier` (e.g. `high`, `middle`, `low`), the runtime picks the corresponding model from the provider
+2. **Provider's middle tier** — if no tier is set, falls back to the provider's "middle" tier (e.g. `claude-sonnet-4-5` for Anthropic)
 
-The default balances cost efficiency with reliable agent behavior. As new models are released, the default may change based on these criteria.
+This ensures cost-efficient defaults while letting experts request more capable models when needed. You can always override the resolved model with `--model` via the CLI or `model =` in `perstack.toml`.
 
 To override the default, specify in `perstack.toml`:
 
diff --git a/packages/runtime/README.md b/packages/runtime/README.md
index fa143de2..da8d4d71 100644
--- a/packages/runtime/README.md
+++ b/packages/runtime/README.md
@@ -18,7 +18,6 @@ import { run } from "@perstack/runtime"
 const checkpoint = await run(
   {
     setting: {
-      model: "claude-sonnet-4-5",
       providerConfig: { providerName: "anthropic", apiKey: "..." },
       jobId: "job-123",
       runId: "run-123",