feat(openai): add GPT-5.5 and GPT-5.5 Pro (#70)

Source: https://openai.com/index/introducing-gpt-5-5/ (announcement 2026-04-23)

Adds 4 new model entries across 3 provider files:

- providers/openai.toml:
    gpt-5.5      — 1M context,  $5/$30 per 1M tokens (input/output)
    gpt-5.5-pro  — 1M context, $30/$180 per 1M tokens
- providers/codex-cli.toml:
    codex-cli/gpt-5.5  — 400K context (Codex subscription limit),
                         $0/$0 (covered by subscription)
- providers/chatgpt.toml:
    gpt-5.5-codex  — 400K context, session-auth (subscription)

Notes:
- API availability announced as 'very soon'; pricing confirmed in the
  announcement. tools/vision/streaming flags match GPT-5.4 family.
- max_output_tokens retained at 128000 (openai/codex-cli) and 65536
  (chatgpt) — matches sibling 5.4 entries; announcement does not
  specify a new output cap.
- Header comment in openai.toml updated (15 → 17 models).

providers/chatgpt.toml

@@ -1,5 +1,5 @@
 # ChatGPT (Session Auth / Codex Responses API) — https://chatgpt.com
-# Models: 5 (free with ChatGPT subscription via OAuth session)
+# Models: 6 (free with ChatGPT subscription via OAuth session)
 
 [provider]
 id = "chatgpt"
@@ -8,6 +8,21 @@ api_key_env = "CHATGPT_SESSION_TOKEN"
 base_url = "https://chatgpt.com/backend-api"
 key_required = true
 
+[[models]]
+id = "gpt-5.5-codex"
+display_name = "GPT-5.5 Codex"
+tier = "frontier"
+# Per OpenAI's GPT-5.5 announcement, Codex exposes a 400K context window
+# for subscription plans (announcement 2026-04-23).
+context_window = 400000
+max_output_tokens = 65536
+input_cost_per_m = 0.0
+output_cost_per_m = 0.0
+supports_tools = true
+supports_vision = false
+supports_streaming = true
+aliases = ["chatgpt-5.5"]
+
 [[models]]
 id = "gpt-5.4-codex"
 display_name = "GPT-5.4 Codex"

providers/codex-cli.toml

@@ -11,6 +11,23 @@ key_required = false
 # gpt-oss-*) are intentionally omitted because the Codex CLI actively
 # migrates users off them via tui/src/model_migration.rs.
 
+[[models]]
+id = "codex-cli/gpt-5.5"
+display_name = "GPT-5.5 (Codex CLI)"
+provider = "codex-cli"
+tier = "frontier"
+# Per OpenAI's GPT-5.5 announcement, Codex exposes a 400K context window
+# (announcement 2026-04-23).
+context_window = 400000
+max_output_tokens = 128000
+input_cost_per_m = 0.0
+output_cost_per_m = 0.0
+supports_tools = true
+supports_vision = true
+supports_streaming = true
+supports_thinking = true
+aliases = ["codex-cli-gpt-5.5"]
+
 [[models]]
 id = "codex-cli/gpt-5.4"
 display_name = "GPT-5.4 (Codex CLI)"

providers/openai.toml

@@ -1,5 +1,5 @@
 # OpenAI — https://openai.com
-# Models: 15 (including 2 Codex variants)
+# Models: 17 (including 2 Codex variants)
 
 [provider]
 id = "openai"
@@ -180,6 +180,35 @@ supports_vision = true
 supports_streaming = true
 aliases = []
 
+# GPT-5.5 — released 2026-04-23.
+# Source: https://openai.com/index/introducing-gpt-5-5/
+# API availability announced as "very soon" at publication; pricing confirmed.
+[[models]]
+id = "gpt-5.5"
+display_name = "GPT-5.5"
+tier = "frontier"
+context_window = 1000000
+max_output_tokens = 128000
+input_cost_per_m = 5.00
+output_cost_per_m = 30.00
+supports_tools = true
+supports_vision = true
+supports_streaming = true
+aliases = []
+
+[[models]]
+id = "gpt-5.5-pro"
+display_name = "GPT-5.5 Pro"
+tier = "frontier"
+context_window = 1000000
+max_output_tokens = 128000
+input_cost_per_m = 30.00
+output_cost_per_m = 180.00
+supports_tools = true
+supports_vision = true
+supports_streaming = true
+aliases = []
+
 # Codex variants (same OpenAI driver & API key)
 
 [[models]]